The Best AI Tools for Designers

• How we got here
• Why AI suddenly became a hot topic
• Chapter Image
• The Best AI Tools
• The Top Ten AI Tools

• 1. ChatGPT
• 2. Gemini – formerly Google Bard
• 3. MidJourney (MJ)
• 4. Dall-E 3
• 5. Adobe Firefly (AF) & Photoshop
• 6. Adobe Express
• 7. Stable Diffusion XL (SDXL )
• 8. RunwayML
• 9. Kaiber
• 10. Topaz
• 11. More Tools Worth Checking-out
• 12. Video demo
• 13. Prompt examples
• 14. New workflows with AI tools
• 15. How to prototype with AI tools
• 16. Case Study – The Real Hero
• 17. Can you enhance your creativity with AI tools?
• 18. Recommendations
• 19. Final Words 

How we got here

Hello Designers, Creatives, and AI Explorers,

Ever wonder how AI is changing the way we work? Do you feel professional pressure to use it but don’t know where to start? Are you feeling anxious about how AI is shaking up the creative communities and taking work away from creators? You may have seen some interesting AI- generated images, and now you feel curious about how to achieve successful results with AI tools.

I feel all these things. With an open mind and a critical eye, I approach this AI exploration, aware of its potential benefits and challenges. My research for this article emerged from a mix of curiosity, excitement, and responsibility. I embarked on a mission to not only understand the technology itself but also explore its ethical implications, limitations, and opportunities for the creative communities to which I belong. I worked for many years as a motion and brand designer for the creative entertainment/ media industries in Los Angeles. I mainly used Adobe Creative Cloud Tools when, in 2022, I discovered AI. My immediate fascination with these emerging technologies propelled me back to the academic world at Hyper Island University, where I wrote my MA thesis on the creative use of Artificial Intelligence. Through my studies and teaching AI workshops with hundreds of participants and many conversations with fellow designers and experts, I hope to share the knowledge and insights I have gained. I tested over a hundred tools and decided to publish the results here.

Three aspects drove my curiosity and motivated me to write this article: 1. How can I maximize designs and creative solutions using AI? 2. The rapid rise of AI and its impact on design. And 3. my passion for lifelong learning and teaching. "AI tools can inspire design, and ambitions are only limited by the imagination," says Tim Brown from design firm IDEO (Fig. 0.1).

Fig. 0.1, Brown, 2022 I am going to share three main points that will change your work! As a designer, I am curious about design theory, history, and methodologies, but I am also a very practical person who wants to get work done. this article's focus is on presenting the best tools and how to use them. Here are the main points that we will dive into:

1. What tools are most effective for text, image, and video creation? Chapters 1–11.

2. What are good examples of prompts and successful results? What areas of work that can be enhanced by AI? The video in chapter 12 demonstrates the results of my AI 12 tool exploration, chapter 12.

3. What are the new workflows you may adopt (Chapter 14)? How do you prototype with clients in a new, more effective, and productive way (Chapter 15)? See how you can improve your work in further recommendations.

These points will be discussed in detail and benefit your know-how, creativity at work, and confidence as a designer in rapidly changing times. AI is here to stay. You're the one in the driver's seat. Written from the perspective of a passionate designer, this article is about defining yourself as a human designer in times of AI. It is about finding a collaborative relationship with AI with realistic expectations and knowing strengths, limitations, and pitfalls. The article unlocks the potential of AI, and helps you to visualize, iterate, and design more creatively and faster. Future-proof yourself with AI tools, boost your workflow, and be creative with confidence. The future is yours to design. Let's get started! Ulrike Kerber Before we dive into creative topics, have you wondered why AI suddenly became a hot topic in 2022?

Why AI suddenly became a hot topic

Why did so many AI tools suddenly pop up? Artificial Intelligence (AI) has been around for years; however, it has become a hot topic and widely debated since the release of ChatGPT in 2022. The maturation of Large Language Models (LLMs) and increased computational power is why so many AI tools suddenly appeared (Shanahan 2023). Doors just opened to the world of powerful language models. Thanks to Application Programming Interfaces (APIs), developers have unprecedented access to these AI tools. This sparks a wave of innovative apps that understand and write text in entirely new ways. These new apps and tools generate images, videos, text, audio, and content, empowering creators and the economy. This leads to a huge economic, creative, and social shift and transformation!

The McKinsey Global Institute projects that generative AI will add the equivalent of $2.6 trillion to $4.4 trillion to the global economy (McKinsey, 2023). However, for working designers, there is a lot of misperception and the fear of being replaced. Misinformation and the negative AI image in the public lead designers to fear and reject AI (Lee, Qiufan, 2023). this article aims to examine the usefulness and creativity of AI tools. Well- timed at the start of AI development, it can stimulate thoughts, approaches, and responsible use before AI "entangles" in everyday life. Inthis early adoption phase, the relationship between humans and AI can be created and defined.

Education, rules, and copyright laws still need to be established in this early adoption phase. The sudden AI tool releases present challenges for creative agencies and designers, who feel pressured to adapt, demonstrate ROI, and future-proof their businesses. Ethical concerns, insufficient expertise, and unclear strategies for AI implementation exacerbate these challenges. Furthermore, creativity is often not a company priority and is not highly valued. The importance of making so-called mistakes, iterating, and mulling over non-finished prototypes in agile work processes is not always acknowledged. Agencies with budget and time constraints think they cannot let designers "play," not realizing that play has an important role in innovation. Experimentation and failures are necessary for innovation. "Outsiders look at Silicon Valley as a success, but it is, in truth, a graveyard. Failure is Silicon Valley's greatest strength. Every failed product or enterprise is a lesson stored in the collective memory. We don't stigmatize failure, we admire it."

– Michael S. Malone

Additionally, there is often a need for creative strategies or innovation frameworks in companies. Agile processes like Scrum or Kanban are widely known now but are often not profoundly embraced or actively implemented. With AI, this may change, as almost anyone (the client, stakeholder, producer, and designer) can create an abundance of professional-looking designs, and prototyping is fast and easy.

Can anybody be a designer now? And what does that mean for the design industry? With hundreds of new AI tools, I present what worked for me: the best, ready-to-use, available tools for design. this article is not about AI programming or training datasets that require deep technical knowledge. this article is a culmination of research, experiments, and conversations with fellow designers and experts. It's an invitation to explore the possibilities and navigate the uncertainties of AI in the creative realm.

The Top Ten AI Tools

The most exciting generative AI tools transform design, communication, and the very nature of artistic expression. As AI rapidly evolves, so do the powerful tools available to you. I've tested over 100 AI tools to discover what ignites creativity most effectively. For a visual demonstration of these tools, please watch the video. However, the vibrant AI landscape is constantly shifting, making it impossible to stay permanently updated. this article will regularly evolve to reflect the latest developments, ensuring you have access to the most valuable resources. While countless niche tools tackle specific tasks, our focus here is on broadbased, market-leading solutions for designers. We'll explore text-to-text, text-to-image, and text-to-video tools, empowering you to push the boundaries of your imagination (Fig. 0.2).

Fig. 0.2, Generative tools, Multiple, 2024

What is the purpose of testing AI tools?

With easy access to Large Language Models (LLMs), app developers have recently been in a race to release the most cutting-edge and innovative tools. This not only results in a crowded market and less-than-optimal tools, but also the danger that some apps released to the public are unsafe. They might have bias, privacy issues, and more problems. You can read more about Ethics in my next article "AI Ethics for Designers". The quality of the tools dramatically varies. While some tools can produce typography, others have trouble with that. There are uneven advantages and disadvantages, and those are good to know because otherwise, you may feel frustrated.

Many digital tools are beginning to integrate AI features; for example, website design platform WIX, however WIX is not entirely an AI tool.

What tools did I examine?

Text Tools: I will start reviewing the two most prominent text-to-text tools, followed by text-to-image and text-to-video tools. I examine text tools because my design process usually begins by writing thoughts and ideas and creating sketches. Writing is essential for any designer to formulate a brand creative brief and a mission statement and to prepare pitches and presentations. It is part of a designer's daily work, and text-to-text tools are essential in the toolbox.

Branding Tools: Some tools, such as Dragonfly, are made for brand designers to analyze graphics with audiences. There are also all-around tools that offer the whole branding process. Unfortunately, these tools try to do too much, and nothing works particularly well. So, as ideal as it may sound to have all tools in one, I would not recommend these (see those tools in the case study). Instead, I suggest combining different tools to forge a personalized workflow for unique results.

As development moves quickly, these all-in-one tools will likely succeed in the near future, and you can have all the tools for branding on one single platform. Many digital tools also have a built-in AI option, such as FigJam, where AI is optional and automatically available on all free Starter teams and paid plans. As mentioned earlier, I will not be looking at digital tools that integrate AI functionality and instead focus on standalone AI tools.

Image and Video Tools: We will dive into a range of available AI tools for image and video creation, catering to various needs and skill levels with pros and cons.

Fig. 0.3, The Author’s description of diffusion models, 2024

How do text-to-image tools work?

Text-to-image tools can be divided into: Diffusion Models, Generative Adversarial Networks, Variational Autoencoders (VAEs) and Transformer-based models.

Diffusion models are currently one of the leading technologies behind AI image generation tools. They work by gradually refining random noise into an image that aligns with a given text description (Fig. 0.3).

Generative Adversarial Networks (GANs) involve two competing neural networks – one creating images, the other evaluating them against real images. Over time, the "creator" network learns to produce increasingly realistic outputs. Examples include StyleGAN and BigGAN.

Advanced designers train their own Generative Adversarial Networks (GANs) by feeding the AI massive amounts of specific data (images, styles, etc.), essentially teaching it to create unique images within that defined style. This involves choosing an appropriate GAN architecture, collecting and preparing data, and iteratively training the model to generate desired results. I have trained StableDiffusion to create an avatar of myself, and the result was a mess of scarry-looking portraits. Either the model or I am not advanced enough.

Variational Autoencoders (VAEs) and Transformer-based models used for language processing and are increasingly applied to image generation tasks. They excel at capturing and using long-range dependencies in text descriptions. Examples include Imagen and DALL–E 2.

How can designers enhance their design practice using AI tools?

The most significant benefit of AI is that you can design prototypes faster and cheaper than ever.

AI tools can:

• Facilitate product or concept Ideation
• Inspire, delight, and surprise through AI art, image and video generation
• Automate Tasks and save time (e.g., retouching or adding styles/filters)
• Create new visual styles. Designers can hone their own skills and perfect their unique styles by combining AI, digital and analog.
• Enhance website design by using AI for agile prototyping, including user testing.
• Analyze data to understand users/ audiences, for example, with a tool like Dragonfly.
• Create multiple variants (colors, layout, or style variations); exemplified

by the ability to visualize seven million different jars of Nutella (Fig. 0.4), Nutella packaging with AI, Dezeen, 2024

The Limitations of AI Tools

Generative AI models are quite powerful and can generate realistic images from text descriptions, but the technology is still under development and has technical limitations. Here are some of the things you should be aware of:

Inaccuracies: Text-to-image tools may not always generate images that accurately match the text description. This is because the model is still learning and may not have the knowledge necessary to understand and generate complex scenes. For example, if you prompt a "dinosaur on green carpet," the model may generate an image of a dinosaur on a carpet, but maybe not a green carpet. This can frustrate users. Text-to-text tools have an error rate, also called hallucination, where the AI goes off the rails trying to please use. Please be careful and do not rely on the accuracy. Research and double-check.

Difficulty with Complex Concepts: Abstract ideas, emotions, and subjective experiences are challenging for AI to translate into visuals. This can lead to generic or clichéd images that fail to capture the desired sophistication and depth.

Quality: Generated images may be non-realistic. This is especially true for complex scenes, objects, or concepts the model has not learned in its training data. For example, if you prompt DALL–E 3 to generate an image of a "dinosaur flying through a city," the model may generate an image that is unrealistic or inconsistent with the laws of physics.

Control and Predictability: While some tools offer control options, achieving the desired image can be difficult. Fine-tuning and iteration are often necessary, requiring time and experimentation. Again, this can be frustrating for the user.

Typography: Creating words, logos, and text can often pose challenges when using AI tools. However, generating letters has become more feasible lately. Successful instances of type and texture generations are explored in the article "Artificial Typography" by A. Trabucco-Campos and M. Azambuja, published by Vernacular (Fig. 0.5). Several designers have experimented with type creations, see Figs. 0.6–0.9.

Fig. 0.5, Artificial Typography Book, 2024

Fig. 0.6, Chensio’s type experiments, 2024

Fig. 0.7, Erik Bernhardsson’s type experiments, 2024

Fig. 0.8, Gianpaolo Tucci’s type experiments, 2024

Fig. 0.9, Puppies and flower’s type with AI, 2024

Bias:

Bias in the algorithm happen, because the models are trained on a massive dataset of images that may contain biases and stereotypes. For example, if you prompt an image of a "world leader," the model will most likely generate a picture of a man. As a result, bias has the potential to detrimentally impact individuals and groups, reinforcing stereotypes, erasing or denigrating them, offering disparately low-quality performance, or subjecting them to indignity. These behaviors mirror biases inherent in the training data for DALL–E 2 and the methodology used in training the model.

For example, the company acknowledged that the explicit content filter applied to DALL–E's pre-training data inadvertently introduced a new bias. "Essentially, the filter—which was designed to reduce the quantity of pre-training data containing nudity, sexual content, hate, violence, and harm— reduced the frequency of the keyword "woman" by 14%. In contrast, the explicit content filter reduced the keyword "man" frequency by only 6%. In other words, OpenAI's attempts to remove explicit material removed enough content representing women that the resulting data set

significantly overrepresented content representing men."(Commonsense Media, 2024).

Technical Limitations: Computational power and training data requirements can limit the resolution, complexity, and diversity of generated images. Some AI apps use so much power, that my computer heats up. Additionally, the cost of some apps can be a barrier.

Size: AI models are limited by the amount of computing power available. As a result, the model can only generate images of a certain size and resolution. However, there are other AI tools specialized in upscaling, see the chapter on Topaz.

Keeping these limitations in mind, I recommend using the tools responsibly and being critical of the generated images. For more ethical considerations, please read my article “AI Ethics for Designers”, coming out soon! Now, let's start reviewing text-to-text tools.

1. ChatGPT

Estimated user base: 180 million monthly users

Price: ChatGPT 3.5 is free, ChatGPT 4 costs $20/month

Capabilities: Text generation in different formats: poems, code, scripts, musical pieces, emails, letters, etc.

Website: https://chat.openai.com

ChatGPT is a chatbot developed by OpenAI in November 2022 and stands out as one of the most renowned and powerful large language models (LLMs). It is available as an Application Programming Interface (API), which means other developers can integrate it into their apps and use the technology for different purposes; see examples below (Fig. 1.1).

ChatGPT stands for "Generative Pre-trained Transformer," which is how it processes requests and formulates responses. ChatGTP's closest competitor, Google’s Bard, is based on another, smaller LLM called LAMBDA. Many more text-to-image tools exist, like Baidu, Microsoft's Turing Models, and Meta’s Opt. However, in this article aimed at visual designers, I focus on the two most prominent text tools useful for design ideation.

Text-to-text tools like ChatGTP perform various tasks, such as answering questions, brainstorming, summarizing content, translating languages, and more (Fig. 1.1). LLMs are the underlying technology that powers text-to-text AI tools. These models are trained on vast amounts of text data to understand and generate human-like text.

Is ChatGPT good at divergent thinking?

I was interested in finding out how designers might enhance their creativity, and divergent thinking is an important component. To put it to a real test, I asked ChatGPT: "Come up with 100 uses for a book." It generated a list in a fraction of a minute (Fig. 1.2). The answers were not very elaborate or unique. It did not consider taking the article apart to create something new in a different medium. I asked the same question again, applying Guilford's measurements, and this time, the results were more inspiring (Fig. 1.3). With hundreds of ChatGPT generated ideas, you can then ask the AI to merge the best ideas into three valid ones. Then you can analyze them or ask ChatGPT to generate the best ideas for your target audience. And if you are not sure about your audience, you can make a customer persona, with demographics and psychographics, also with ChatGPT. The possibilities are endless.

Fig. 1.2, ChatGPT 4, Divergent thinking answers, 2023

Fig. 1.3, ChatGPT 4, 2023

What creative things can ChatGTP do?

Fig. 1.4, The Author, inspired by “The Coming Wave, Suleyman, 2023

The Positives

ChatGPT

Boosts Creativity: ChatGPT enhances creativity by providing creative responses, making it a valuable tool for ideation.

Is Versatile: ChatGPT is adaptable and can be employed for a wide array of assignments and tasks (Fig. 1.4).

Enhances Productivity: Writing tasks are accomplished more efficiently, including content creation, rephrasing, generating variations, and adapting the tone for different audiences and purposes. For instance, web developers who would typically spend hours writing lines of code for a website can now utilize ChatGPT to complete the task in a fraction of the time.

Supports Learning: ChatGPT can help you hone your skills. You can learn a lot about, for example, fiction or marketing writing using Chat GPT. Not only do your writing skills become better, but your knowledge level improves. Always remember to fact-check!

It improves: By upvoting or downvoting a particular response, your feedback provides further improvement with the release of newer versions.

Assists research: As an example, if you are designing a website for selling handbags, you could provide a prompt like "Create a customer persona who purchases eco-friendly luxury handbags" to gain a better understanding of this target audience. However, it is crucial to verify and corroborate this data using various sources before relying on it for decision-making.

The Negatives

Errors/ Hallucinations: One of the most serious and potentially dangerous disadvantages of ChatGPT is that it produces inaccurate or incomprehensible texts in the midst of generating plausible and compelling responses. The text sounds professional, and one could easily assume that the content is correct, which is misleading. I constantly have to remind myself that the test has a 15–20 percent hallucination rate and that I should not trust it, however persuasive it sounds.

Impact on Job security: The assistance provided by ChatGPT in tasks such as writing social media content, articles, and books raises concerns about job security, potentially leading to job losses before new opportunities emerge. This leads to a widespread transformation that will affect many.

Generic: ChatGPT’s answers are generic and hollow.

Outdated Information: Some limitations might exist depending on the specific features and subscription options used. For example, the free version of ChatGPT might have less access to real-time information compared to the paid version.

Privacy Concerns: ChatGPT retains and stores all conversations as chat history to enhance its language model. Your chats may be seen by human trainers working behind the scenes at OpenAI. (Tech.co, 2024). One solution is to download a large language model and run it on your own machine. This would ensure that an outside company would not have access to your data. (Infoworld.com, 2024).

How to use ChatGPT – Prompting Rules

There are numerous articles with excellent examples and tips for writing prompts in the areas of marketing, advertisement, customer service, sales, user interface, e-commerce, surveys, instructional learning, workflows, methodologies, and more (see sites like Fig.1.5 from Expand or Semrush).

There are also prompting assistants, as used in the case study. Here are some tips:

Provide relevant details: Specific details help the platform understand the specific scenario or context. It can provide more accurate responses with background info, specific facts, or user preferences (Fig.1.5).

Assign roles: Make sure you identify what role ChatGPT should provide (You are a university professor who assesses, and I am a student of digital management). Assigning a specific role to ChatGPT can help guide the responses and align with the desired expertise or perspective.

Edit: You can edit an original prompt by hovering over it and clicking the edit button.

Safeguard personal information: Don't feed ChatGPT any sensitive personal or company information you don't want available to anyone else. OpenAI stores queries: "Your conversations may be reviewed by our AI trainers to improve our systems." (OpenAI, 2024). The more data is stored online, the more risk of getting hacked (Semrush, 2024).

Fig. 1.5, Expand, Prompting Rules, 2023

Fig. 1.6, The Author Rule of thumb, 2024

Ideation sprints with ChatGPT or other text-to-image tools

Often, traditional brainstorming sessions don't work, because people have a bias toward agreement and conformity. A promising idea early in a brainstorm tends to shape the rest of the session as people gradually fall in line. Second, participants tend to lose their train of thought as other people speak aloud or share in front of the group. (Dropbox Blog, 2024).

ChatGPT does not lose a train of thought and is, in a sense, naive and not cultured, which is an advantage when it comes to coming up with fresh ideas.

Great ideas are built over time, and there is no magic bullet. But if you like to speed up the ideation phase, AI can assist. You can ask ChatGPT: What would be the next product in my line of products? What services should we offer in the future?

Great Industry and Market Trends Prompts:

• 10x Thinking: Imagine the world 10 years from now. How will your industry have changed dramatically? What threats and opportunities will emerge?
• SWOT Analysis: Analyze the Strengths, Weaknesses, Opportunities, and Threats facing your business within your current and future market landscape.
• Competitor Analysis: How are your competitors preparing for the future? What lessons can you learn from their strategies?
• Emerging Technologies: Identify 3–5 emerging technologies with the potential to disrupt your industry. How can you leverage them to your advantage?

You can future-proof yourself with a ChatGPT career consultation: What are my next career steps? In that way, you can view ChatGPT as a buddy, a non-biased friend with whom you can collaborate (well, it also has biases, but different than your biases).

Career Building Prompts:

• Portfolio development: How can you build a portfolio showcasing your skills and experience, even if you're not changing careers right now?
• Future: Imagine you have a crystal ball and can see your future self 10 years from now. What does your ideal work life look like?
• Networking and personal branding: How can you strengthen your network and build a strong personal brand to stay relevant in the job market?
• Lifelong learning: What resources and strategies can you use to continuously learn and upskill yourself throughout your career?
• Impact and purpose: What kind of impact do you want to make with your work? How can you align your career with your purpose?

With AI, sprints are better, more productive, and faster. Often done in the UX field, a design sprint is to build and test a prototype in just five days. With a small team, you rapidly progress from problem to tested solution using a proven step-by-step checklist (Fig. 1.7). You can see how customers react before you invest all the time and expense into creating your new product, service, marketing campaign... or whatever! (The Sprint Book, 2024) Fig. 1.7, The Sprint Book, 2024

Conclusion

ChatGTP helps with countless creative tasks (Fig. 1.4). It is a great brainstorming partner but not necessarily a source of truth. AI's full potential is yet to be explored. Prompting is a skill with rules (Fig. 1.5). AI offers creative solutions only if prompted correctly (Fig. 1.5). With better, more specific prompts you get better results. However, even mediocre answers or "mistakes" can inspire the designer. I recommend you to experiment with various text-to-text tools, as different tools may be more suitable for specific purposes. For a comparison, please go to the next chapter. Text-to-text AI fundamentally changes the way we think, learn, educate, live, and work. According to innovator Brian Mattimore, an ideation sprint that usually takes 5 days can be sped up to 2 days using ChatGPT. In 10 minutes, you can generate 5000 ideas, which typically take all week without AI (Mattimore, 2024). This process augments innovation. AI is naive and without (your) bias (it has other biases than you). It can, therefore, detect your assumptions and help you get a fresh view. However, text-to-text tools must be implemented carefully and consciously to avoid creating and sharing misinformation. Let's look at another valuable tool in your toolbox: Gemini or formerly Google Bard.

2. Gemini – formerly Google Bard

Estimated user base: 142.4 million/month

Price: free

Capabilities: Text Generation in different text formats: poems, code, scripts, musical pieces, emails, letters, etc

Website: https://bard.google.com/chat

Google's AI chatbot Bard has changed its name to Gemini. While the interface boasts some improvements, you can still access your saved chat history from Bard seamlessly. The language model powering the new Gemini chatbot is Gemini Pro, which replaced PaLM 2, the language model that has powered Bard previously, in December of last year. Before that, it was running on LaMDA, the same model that Google once infamously fired an employee for claiming was sentient. Gemini is based on one of the biggest Large Language Models (LLMs) and is a competitor to ChatGPT. I asked Gemini the same question “Come up with 100 uses for a book” (the alternative-use test) and to answer it as creatively as possible using Guilford’s four criteria for creativity. Here are some inspiring results (Fig. 2.1):

Fig. 2.1, The Author, Google Bard, 2024

Testing Gemini

To make this test applicable to a design project, I asked Gemini to develop ideas for an imaginary advertisement for sneakers using Guilford's Alternative Use Test and make it as creative as possible (Fig. 2.2). Some of the answers are fun and original.

Fig. 2.2, The Author, Google Bard, 2024

You can take the Pros and Cons from the ChatGPT chapter since both platforms offer the same. However, here are a few additions:

The Positives

• Integration with Google products: May integrate seamlessly with other Google products you already use, streamlining workflow.
• Free tier available: Offers a free tier with limited usage, allowing you to experiment and test the tool before committing.

The Negatives:

• Limited control and transparency: The inner workings of the model are not fully transparent, making it difficult to understand how it generates outputs and potentially leading to unexpected results.
• Limited explainability: It can be challenging to understand the rationale behind the outputs generated by Gemini, making it difficult to assess their accuracy and reliability.
• Limited Availability: Image generation in Gemini Apps is available in most countries, except in the European Economic Area (EEA), Switzerland, and the UK. It’s only available for English prompts. (Google, 2024)

So, what is better, ChatGPT or Gemini?

ChatGPT–4 and Gemini are similar, but powered differently. ChatGPT–4 is powered by the Generative Pre-trained Transformer 4, while Bard is powered by the Language Model for Dialogue Applications (LaMDA). GPT–4‘s LLM can generate text, translate languages, write different kinds of creative content, and answer questions in an informative way. See a comparison of results: Fig. 2.5.

According to Tech.co Gemini exhibited improved content ideation capabilities on the 2024 tests. Gemini's responses were often set out in a more readable format than ChatGPT's responses. ChatGPT was better at paraphrasing, but Gemini was better at simplification (Fig. 2.4).

Fig. 2.4, Tech.co, ChatGPT and Gemini in comparison, 2024

Fig. 2.5, The Author, ChatGPT and Bard comparison, 2024

Conclusion

ChatGPT is efficient at generating and summarizing text requests and is good at paraphrasing. Gemini is better at answering questions with more relevant, denser information. I asked both AI tools the same questions and received very different answers (Fig. 2.5). It is difficult to determine which is “better.” ChatGPT responses are typically broad and lack a specific tone (somewhat wishy-washy), while Gemini offers a more defined perspective. I suggest using both and evaluating the outcomes based on the project's objectives.

Google is constantly releasing new models, like Gemma, a family of lightweight, open models built from the research and technology that Google used to create the Gemini models. These are worth checking out, once they are fully released, since Google has a strong research team and offers some of the best solutions.

In addition, I suggest adding a text-to-text tool like Perplexity.ai that shows the internet sources right away so that the research can continue more deeply with more clarity and accuracy.

Let’s look at Image creation tools next!

3. MidJourney (MJ)

Estimated user base: 50,000+

Prices: $10 Basic, $30 Standard, $60 Pro, $120 Mega

Capabilities: text-to-image, image-to-image, light image editing

Website: https://www.midjourney.com

Handbook: https://docs.midjourney.com/docs/quick-start

MJ is an image generation tool that takes text prompts and uses a machine learning algorithm trained on a large amount of image data to produce images. It draws from over 1 billion images, photographs, art, and illustrations from the internet (without asking for the rights). Its image generation algorithm is trained on language and diffusion models (Artnews, 2023). MJ excels at producing immersive environments, particularly fantasy and sci-fi, with dramatic lighting that looks like rendered concept art. The generation process happens through the gaming platform Discord. Through much testing and comparison, I find MJ’s aesthetic, creativity, and elaboration superior to the competitors in many aspects (Fig. 3.1 and 3.2). The texture of the sneaker is more defined, and the colors are more vivid. The MJ result looks like a professional higher-end photo shoot, while the competitor’s results look amateur.

Fig. 3.1, The Author, image comparisons, 2024

Fig. 3.2, Tool comparison, ML news, 2023

What is a MJ prompt?

A prompt is usually a text that the Midjourney Bot interprets to produce an image. A prompt starts with /imagine and a description which can be a single word, a sentence, an emoji, or an image. Advanced prompts include options and parameters (Fig. 3.3).

Fig. 3.3, Midjourney Handbook, 2023

The Positives:

Artistic: MJ generates high-quality, well-defined, versatile, and artistically diverse images. The latest version generates photorealistic images that even deceive experts. In a photography contest, an MJ-generated image won, demonstrating how far AI image generation has come (Fig. 3.4). New MJ system versions continuously improve efficiency, coherency, and quality (MJ, 2023). My MJ portraits since 2022 show the algorithm’s progress with improved results over one year (Fig. 3.5 and 3.6).

Fast: Users can iteratively improve images and create revisions in a fraction of the time it would take a designer. Getting started with a comprehensive Handbook and supportive community is easy.

Quality: The quality of the images can be top tier.

Features: MJ is versatile and full of features, from remixing and editing specific portions of an image with the 'Vary Region' feature to the flexibility offered by its zoom and pan capabilities.

Fig. 3.4, Absolutely AI, Redshark news, 2023

Fig. 3.5, The Author, MJ Version Comparison, 2023

The Negatives:

Details: Portraits can look lifeless since the eyes do not have highlights. The algorithm often fails to generate details like eyes, fingers, and hands.

Prompt writing: there is a steep learning curve. It lacks logic and clarity and can get complicated and time-consuming quickly. The more concrete the desired outcome, the more frustrating it can be.

Idealized images: It requires experience to detect the difference between a real photo and a too-beautiful, idealized MJ-generated image (Fig. 3.4).

Discord: MJ can be accessed through Discord, which has a unique interface. While avid Discord users may navigate it seamlessly, newcomers might find it somewhat overwhelming. Privacy concerns related to Discord keep users cautious (Makeuseof, 2023).

Copyright: Images may be deceiving since no watermark says the image was AI-generated.

Price: MJ initially offered a free trial, but now operates on various pricing plans, making it a consideration for users evaluating cost-effectiveness (Reddit, 2023).

How creative can Midjourney get?

To achieve high scores on Guilford’s creativity scale plus “Sparkle” (my personal addition to measuring creativity). I have focused on more advanced features: 1. multi-prompt, 2. remix, 3. prompt assistants, 4. blend, and 5. image prompts with textures. Those are all well-described in the MJ Handbook (Fig. 3.7).

1. Multiprompt

The most creative results are achieved through complex prompts juxtaposing different domains, referencing architecture, fashion, art, and science. Styles differ from other versions (Fig. 3.6).

Fig. 3.6, The Author, multi-prompt, 2023

Fig. 3.7, Midjourney Handbook, 2023

Multi-prompting was tested for a “Carfish” transportation hybrid that I tested for a workshop about future vehicles (Fig. 3.8).

Fig. 3.8, The Author, Multi-prompt example, 2023

Fig. 3.9, The Author, Remix Example, 2024

2. Remix

From four results, the user can upscale, create variations, or add to the prompt and change parameters with the remix button. I chose image 4 to remix (Fig.3.8), resulting in Fig. 3.9.

The resulting images look more like a future hybrid vehicle.

3. Prompt Assistants

ChatGPT can be used as a prompt assistant. For “Carfish” transportation hybrid, it wrote this prompt. The result is a detailed, futuristic world with transportation devices close to what I was aiming for (Fig. 3.10).

Fig. 3.10, The Author, MJ prompted by ChatGPT 4, 2023

To enhance creativity even more, I experimented with another prompt assistant that offers a multitude of options for customizing images: Promptfolder.com (Fig. 3.11).

Fig. 3.11, The Author, Promptfolder, 2023

Promptfolder suggested (Fig. 3.12):

Fig. 3.12, The Author, MJ with Promptfolder, 2023

The result is a surreal, detailed transportation vehicle (Fig. 3.12). The image could be used for animation storyboards. To evaluate creativity, I would give this image:

1. Fluency – 10
2. Originality –10
3. Flexibility – 8
4. Elaboration – 9
5. Sparkle: 5

(1= low, 10= high)

4. Blend

Using the Blend function, two or more images are integrated into one. This blending would be highly time-consuming or impossible to do traditionally or digitally. “Blend” results came as a positive surprise. They can be very creative when choosing the right images that complement each other (Fig. 3.13, 3.14). They can lead to prototypes and new product designs.

Fig. 3.13, The Author, MJ Blend, 2023

Fig. 3.14, The Author, MJ Blend, 2023

5. Prompting with an Image

Successful results can also be achieved by using an image as a reference. In this example, Chris Branch used textures and materials to create images of smart phone covers (Fig. 3.16).

Fig. 3.16, Artificial Inspiration, LinkedIn, 2024

Origami, felt, wood, velvet, and silk, are some of the most common textures/materials used in prompts (Fig. 3.17).

Here is a list of textures to be added to your library of prompts:

• Natural textures: Bark, moss, lichen, sand, stone (specify type like slate, marble, etc.), wood (grain type, weathered, polished).
• Fabrics: Satin, silk, brocade, denim, tapestry, macrame, knit, crochet, woven, ripped, patched
• Metals: Brushed steel, hammered copper, oxidized bronze, tarnished silver, cast iron, damascus steel, filigree.
• Other materials: Clay, ceramics, pottery, glass (stained, etched, blown), paper (parchment, papyrus, corrugated), pearls, mother of pearl, leather, fur, feathers, scales, bone, ice, crystals.

Descriptive Adjectives:

• Include adjectives to convey texture: Rough, smooth, bumpy, pitted, cracked, weathered, worn, aged, polished, glistening, iridescent, translucent, shimmer upcycled.
• Combine textures with visual metaphors: "Clouds like whipped cream," "Skin like polished marble," "Hair like spun gold."

Artistic References:

• Mention specific artists or styles known for their use of texture: Van Gogh's impasto brushstrokes, Monet's water lilies, Klimt's gold mosaics.
• Reference art movements based on texture: Impressionism, Pointillism, Art Deco.

Fig. 3.17, Chris Branch, Linkedin, 2024

You can upload your logo to MJ if you want to achieve typography with origami paper texture. In this case (Fig. 3.18), I have uploaded several jpgs of the “viva” letters in different typefaces. Then, I used them as a reference to create origami letters.

Fig. 3.18, The Author’s logos created with MJ, 2024

Advanced prompts include a subject, references ( known artists) and descriptions (Fig. 3.19). In Version 6, you can write complete sentences, not snippets divided by commas.

Fig. 3.19, prompt by stella sky, 2024

A few more prompting rules for MJ version 6:

• For incorporating texts use phrases like: says, printed on, entitled, inscribed with, labeled as, marked with, branded with, embossed with, engraved with, stamped with, adorned with, scripted with, lettered with , etc.
• Communicate with Version 6 using clear and simple language and correct grammar and punctuation to avoid miscommunication
• Do not use exaggerated prompts, words like ultra, super, hyper, insanely, extremely, quite, rather, somewhat, notably, especially, significantly, remarkably
• You can also print text on things, such speech bubble, post-it note, article cover, poster, sign, t-shirt, mug, billboard, newspaper, magazine, greeting card, envelope, license plate, calendar, ticket, product packaging, business card, etc.
• To have text or letters appear alone, it sometimes helps to add the phrase typography design to the prompt. You can explore using phrases isolated on a white background if you’d like the canvas to be otherwise blank (UXBootcamp, 2024).

Fig. 3.15, The Author, Prompt: blue butterfly, Menelaus blue morpho, close-up, photorealistic on white background, 2023

Conclusion:

MJ proves to be an excellent tool for prototyping as long as the user aims for an unexpected, somewhat random result. The more advanced the prompt, the more creative the result.

What can you do with results that are not entirely finished or complete? Logo results (Fig. 3.18) can serve as first ideation for your client. Once the direction is approved, I would either create a 3d model of the origami logo for added flexibility, upscale it with Topaz or rework it in Photoshop.

MJ also generates successful results when the prompt is simple and specific (see blue butterfly, Fig. 3.15).

The algorithm does not consistently take into account all components of the prompt. Due to the difficulty of manipulating details and the lack of control over results, text-to-image tools are not yet fully reliable as design tools. That is why many designers, like myself, are not using the images for clients yet. This will change. Since I started writing this article in the early summer of 2023, AI has already improved significantly. Adobe has introduced many new AI features. As AI continues its evolution, it will soon possess the capability to generate logos, letters, numbers, and more elaborate details and offer increased control over the results.

Let’s move on to MidJourney’s competitor Dall-E 3.

4. Dall-E 3

Estimated user base: One million/month+

Price: Free with ChatGPT 4 ($20/month) and also free for a limited amount of images (at time of writing 20 images)

Capabilities: text-to-image, image-to-image

Website: https://openai.com/dall-e-3

DALL–E 3 is the latest iteration of OpenAI’s model, specializing in text-to-image generation. When you open ChatGPT 4, you can just type “Make an image of …, which is easy and convenient. It produces a wide range of images based on short or detailed descriptions, covering everything from basic objects to intricate scenes and abstract ideas. It understands text well and generates images to produce various visual content. It can craft movie posters, visualize storyboard scenes and architectural designs, conceptualize fashion ideas, prototype products, and more.

Fig. 4.1, The Author, Upcycled sneaker to a garden pot, MidJourney and Dall-E3, 2024

The Positives:

Practical: Dall-E3 is embedded into ChatGPT. It has a simple interface, is swift, and is user-friendly.

Type: It is better than MidJourney when it comes to type generation.

Details: It is also better at generating fingers and other details than MJ. It is pretty photo-realistic.

Improves: It continuously improves images through conversational feedback.

The Negatives:

Limited styles: Dall-E3-generated images look straightforward and lack the charm or emotional engagement of MidJourney’s generations; see examples below.

Style: The style are not photorealistic but illustrative (see a portrait of a woman in Fig. 4.4). Not surprisingly, it generates younger-looking women; apparently, older women are not beautiful.

Bias: Just from ordinary photo prompts, the AI was far more biased against female subjects versus male. Age and race bias are prevalent.

Details: It still has hands and fingers issues.

Comparison MidJourney and Dall-E3

On this page, I show various image comparisons between MidJourney and Dall-E3 test results using the same prompts (Fig. 4.1 to 4.4). For the style and look that you want, I encourage you to do the same and try several platforms to achieve the look that you want.

Prompt: Ad poster for a blend of running shoe and high heel. MJ did not create a high heel and has trouble with type and logos, Dall-E3 is better at creating type.

Fig. 4.2, The Author, MJ and Dall-E3

Fig. 4.3, Donavan Rittenbach’s comparisons, Medium, 2024

Fig. 4.4, The Author, MidJourney and Dall-E 3 prompt: image of a beautiful, natural woman's face, created in a photorealistic style, 2024

Conclusion

For straightforward visuals for prototypes, where sophisticated, photorealistic styles are less important, Dall-E3 is a practical and reasonable option (Fig. 4.5). For more sophistication, other platforms like MidJourney are preferred. Let’s examine a very user-friendly platform next: Adobe Firefly.

Fig. 4.5, The Author, Dalle E, 2024

5. Adobe Firefly (AF) & Photoshop

Estimated user base: NA

Price: free and premium plans

Capabilities: Text-to-image, generative fill, generative expand, moving elements, text effects. recolor vector graphics, text-to-vectorgraphic

Website: https://firefly.adobe.com

Handbook and Support: https://helpx.adobe.com/support/firefly.html

What makes Adobe Firefly (AF) different from competitors? According to Adobe, Firefly follows the Adobe AI ethics principles of accountability, responsibility, and transparency. "Feel at ease knowing you're creating AI images from data based on Adobe Stock, openly-licensed content, and public domain content where the copyright has expired." (Adobe, 2024)

AF is trained on hundreds of millions of Adobe Stock images, openly licensed content, and public domain content where copyright has expired. Creators who opt out will be excluded from training data sets, however there is still concern that this isn't an explicit opt-in consent instead. (1827 Marketing, 2024)

Also, a big difference to other models is the vector graphic capability: "… the world's first generative AI model for vector graphics and first generative AI model to generate “human quality” vector and pattern outputs." (Adobe, 2024)

Settings

Adobe Firefly is easy and user-friendly with the same vocabulary as other Adobe products. It offers the following options, making it highly convenient for beginners (Fig. 5.1):

1. Text-to-image
2. Generative Fill
3. Text effects
4. Recolor (and soon others such as 3d to image, and objects)

Fig. 5.1, Adobe Firefly, 2024

Option 1. Text-to-image

This function is similar to other AI image creators, except there are parameters to choose from to create a style, even with uploaded reference images (Fig. 5.2).

Fig. 5.2, Adobe Firefly, 2024

Option 2. Generative Fill

With Generative Fill, backgrounds can be removed, altered, or extended—a very practical tool, just as the Photoshop AI capabilities (Fig. 5.3).

Fig. 5.3, The Author, Adobe Firefly, 2024

Option 3. Text Effects

There are quite a few text effects that can be fun but also kitschy or a bit "too much". However, it is useful to have these effects for certain projects (Fig. 5.4).

Fig. 5.4, The Author, Adobe Firefly Type, 2024

Option 4. Re-color

Vector images can be uploaded and recolored in a second, a real time-saver!

Fig. 5.5, The Author, Adobe Firefly, 2024

The Positives:

Text: AF is better at generating letters and readable words where other platforms struggle. Firefly has a good text generator to create typography.

Adobe: It can be convenient to stay in the "Adobe family" when already using other Adobe products.

Price: Currently in an open beta phase, it offers free access but may require admission into the beta program for full features.

Reference Options: It offers a reference library and the option to upload your style to choose the desired art direction.

Generative Fill: A feature that allows users to overlay additional elements onto images, enhancing the creative process.

Text Effects and Generative Recolor: These features cater specifically to the demands of designers, allowing for significant time savings and creative experimentation (Fig. 5.5).

Authentic: AI image generators can create striking fake images that seem real and can be misleading. To tackle this, Firefly uses Content Authenticity Initiative (CAI) metadata to fingerprint any images it generates.

Copyright: Firefly is commercially safe, as it is trained on licensed content, such as Adobe Stock, and public domain content where copyright has expired.

You can track the origin and history of AI-generated elements within your work.

The Negatives:

Looks like stock photos: Adobe Firefly draws from stock photos therefore, results are idealized and too “beautiful” or perfect and cookie cutter.

Less Control: Adobe Firefly gives users less control over their designs.

Not customizable: Users can influence the style only with templates or uploadable images. This makes the designs look more homogenized.

Limited: It does not generate celebrities or brand names. I typed in "Adidas" Sneaker and it took the word Adidas out. Fig. 5.6 does not resemble an Adidas Sneaker.

Fig. 5.6, The Author, Upcycled Adidas sneaker to a garden pot, Adobe Firefly, 2024

Photoshop

Firefly is available in Photoshop (try Generative Fill and Generative Expand, also in your browser with Photoshop on the web), Illustrator (try Generative Recolor), Adobe Express (try Text to Image and Text Effects) and as a standalone web app at firefly.adobe.com. (Adobe, 2024)

Photoshop AI is based on Firefly but works a bit towards the designer used to working with Photoshop’s interface.

Adobe Firefly offers beginner-friendly AI image creation with "ethical" data, potential for seamless Adobe workflow, and commercially safe outputs, but it's newer (less experienced than other models). Adobe Express, is another Adobe option, see next.

6. Adobe Express

Estimated user base: NA

Price: Free, includes thousands of templates, Adobe Stock photos, videos and music, fonts, design assets, quick actions, and amazing features. The $9.99/month Premium Plan unlocks all content.

Capabilities: text-to-image, image-to-image, vector graphics, image editing, and video creation

Website: https://new.express.adobe.com

Handbook and Tutorials: https://helpx.adobe.com/express/user-guide.html

Adobe Express is similar to Canva in that it is geared towards design beginners, offering a template-based design tool for the quick and easy creation of content. Adobe Express is revolutionizing how people and brands around the world turn ideas into standout content with groundbreaking generative AI designed to be safe for commercial use, coupled with an easy-to-use yet powerful all-in-one editor for all content creation needs (Adobe, 2024).

The Positives:

User-friendly: The tool is handy for beginners and self-proclaimed non-creatives. It is similar to Canva, used by many designers.

Templates: There are a lot of templates to choose from, eliminating the need to start from scratch.

Fast: It is designed for practical applications such as quickly creating an Instagram post/story, a flyer, a TikTok video, a logo, a poster, a video, and similar. You can generate an image, remove the background of an image, resize, remove, or convert to JPG or PNG. This app definitely makes it easier for designers to do daily tasks like creating social media content in a short amount of time.

Save Money: This tool helps individuals who are not trained in graphic design, but who are creative and possess a good design aesthetic to bypass the need to hire a graphic designer.

Copyright: While the Adobe Express platform itself is not copyright-free, the stock assets provided within the platform are generally covered by royalty-free licenses, allowing you to use them in your projects without paying additional fees. Some Adobe stock photos come with a premium badge are provided under a Limited License.

Teams: Teams can collaborate in real-time via the commenting function.

The Negatives:

Replaces freelance designers: Because it is such a user-friendly, easy, and effective tool, it may also replace designers because untrained individuals can dive into image creation without training and generate professional-looking results.

Homogenized design: While tools such as Canva or Adobe Express offer convenience, the results can be overly standardized, resulting in a generic appearance where everything starts to look alike—especially if Adobe is not actively addressing this issue through frequent updates. This could potentially diminish the uniqueness, professionalism, and sophistication of designs. However, the impact of this concern is still unfolding in the dynamic landscape of AI.

Limitation: You cannot use celebrities or brand names as references.

Lack of Control: One key disadvantage of Adobe Express/ Firefly is limited user control over the final design. It offers template-style features (Fig. 6.1).

Lack of Fine-tuning: While you can provide prompts and choose from different variations, you can't fine-tune specific aspects like layout, color scheme, or typography. This might be frustrating for experienced designers who require precise control over their creations.

Conclusion

As a designer, it's crucial to refine your style and carve your own path by utilizing various tools, experimenting with new techniques, and combining styles until you discover your unique artistic expression. Cultivating this individuality is key to positioning yourself as a valuable asset in the marketplace. As a designer, you can read trends quickly and respond to the ever-changing environment in a way that AI platforms cannot. This gives us human designers an edge over the overwhelming AI presence in all creative domains. Canva or Adobe Express offer cookie-cutter solutions and are not offering brilliant solutions but middle-of-the-line images. Those could be used as prototypes in the process or are valuable assets for low-budget solutions.

I would use Adobe Express for ideation and prototyping when I have little time or a small budget. Let’s look at Stable Diffusion next for a more in-depth AI model.

Fig. 6.1, The Author, Adobe Express features, 2024

7. Stable Diffusion XL (SDXL)

Estimated user base: 10 million+

Price: Free. More features with subscription plans: Basic plan is $9/month, standard plan $49/month, premium plan $149 per month

Capabilities: text-to-image, image-to-image, graphic artwork, image editing, text-to-audio, clean audio, remove silence, transcript, subtitles, training AI models, 3d models and video creation

Website: https://stability.ai

Stable Diffusion XL (SDXL) is a text-to-image model by StabilityAI. It is a deep learning, text-to-image model released in 2022 based on diffusion techniques. It is primarily used to generate detailed images based on text descriptions. However, it can also be applied to tasks such as in-and outpainting, and generating image-to-image translations guided by a text prompt. Its development involved researchers from the CompVis Group at Ludwig Maximilian University of Munich and Runway with a computational donation by Stability AI and training data from non-profit organizations.

Stability AI's First Open Video Model is open, and I joined the waitlist. The differences between models is described here: Fig. 7.1. Stable Video is designed to create videos for media, entertainment, education, and marketing.

Fig. 7.1, Different Stable Diffusion versions, 2024

Features:

Stable Diffusion XL represents a major upgrade over previous versions of Stable Diffusion, offering several new capabilities and improvements that take AI image generation to the next level: Higher image resolutions, better photorealism, enhanced text generation (Fig. 7.3).

Inpainting, outpainting, and image-to-image generation give artists flexibility and control over the visual synthesis. Just like in Photoshop AI, you can:

• Remove unwanted objects from photos and fill in the blanks in a coherent manner
• Extend the boundaries of an image naturally, continuing patterns and textures
• Modify an existing image by modifying the prompt while keeping the general composition (Magai, 2024)

Fig. 7.3, The Author, Stable Diffusion features, 2024

How does Stable Diffusion work?

Unlike many other image generation models, Stable Diffusion uses a diffusion process with Gaussian noise to encode and recreate images. What sets it apart is that it operates in a reduced-definition latent space instead of the pixel space of the image. This choice is driven by the significant reduction in processing requirements (Stable Diffusion, 2024).

The Positives:

Environmentally more sustainable: A key advantage of Stable LM 3B is its smaller size and efficiency. Unlike larger ones, these models require fewer resources and have lower operating costs, making them highly accessible for most users. Not only does this make them more affordable, but it also makes them environmentally friendly, as they consume far less power. (Stability.ai, 2024). This is something that Stability.ai claims and is hard for me to prove.

Accessible and easy to use: It is a professional tool but can run on consumer-grade graphics cards.

Control: You control key hyperparameters, such as the number of denoising steps and the degree of noise applied (amazon, 2024). Also, you can change the prompt while it is generating. I added people around the Adidas Sneaker Statue (Fig. 7.4).

Reliable: the technology guarantees long-lasting durability and reliable performance. This means developers can expect their Stable Diffusion models to maintain effectiveness despite constant use.

Training your own: You can train your own AI, for example, to make your own avatar. You upload photos of yourself, and the AI generates countless versions of your image in illustrative or photorealistic styles.

3D object generation is a great feature, but it comes only with the membership. Motion designers will soon be able to upload an image and have Stability AI create the 3D Asset. This is huge.

Fig. 7.2, The Author, Stable Diffusion prompt: Astronaut in a field of flowers, 2024

The Negatives:

Bias: Just like all other AI models, there is bias to be aware of. In Fig. 7.2: The American flag or the USA were not mentioned in the prompt. The prompt just said "astronaut in a field of flowers".

Harmful images: The images Stable Diffusion was trained on have been filtered without human input, leading to some harmful images and large amounts of private and sensitive information appearing in the training data.

Complex interface: requires learning time.

Time-consuming: Like with all the AI models, micromanagement is necessary to achieve desired results, which can be time-consuming.

Aesthetics: Like Dall-E3, Stable Diffusion is straightforward (Fig.7.2 and 7.4) and less sophisticated or engaging.

Limitations: Training custom AI models sounds exciting, but when I tried to generate my own avatar, the result was quite disturbing because the images did not resemble me at all (and I am better not showing the result here:)

Fig. 7.4, The Author using Stable Diffusion Prompt: adidas sneaker made out of glitter and pearls in a giant size as a statue in a public square with people around it, 2024

Conclusion

Stable Diffusion offers many features, handy for motion designers with audio and video-generating skills. It compares to RunwayML, which appears to be more user-friendly. RunwayML is preferred by some video creators (Reddit, 2024); see next chapter.

8. RunwayML

Estimated user base: 10,000+

Price: Free. More features with $12 basic plan, $28 Pro, $76 Unlimited

Capabilities: text-to-image, image-to-image, graphic artwork, image editing, text-to-audio, clean audio, remove silence, transcript, subtitles, training AI models, 3d models and video creation

Website: https://runwayml.com

RunwayML is a comprehensive, intuitive, cloud-based platform with various tools for artists without coding experience for media ranging from video, audio, and voice to text. I trained the model to create my voice as an AI voice (Fig.8.1).

Fig. 8.1, Ulrike Kerber's voice as an AI voice using RunwayML training, 2024

The Positives:

Custom training: You can train your own AI model, as I did with my own voice (Fig. 8.1).

Generation features: You can prompt with a text or image and generate images or videos. You can customize images and adjust colors, styles, and compositions. The dashboard is easy and intuitive.

You can also upload a video as a prompt, and the motion of the video will remain, but the texture, color, etc. can be changed. You can remove and replace the background of a video. Gen-2 offers over 30 variations of tools. Text-to-color grade is an interesting feature that lets you experiment with the colorization of the video and give it a new style.

Other more useful features: Runway's Clean Audio, Transcript Tool, Subtitle Tool, Silence Remover Tool.

Motion Brush is a feature that lets you isolate the part that you want to move.

Audio: The audio feature is useful for any motion designer/video creator: Text-to-speech, clean audio, removing silence, transcribing, and Subtitles.

Other functions worth mentioning: Sync audio and video, removing background noise, and more for video creators.

Styles: For image creation, there are a lot of styles to choose from (Fig. 8.2) and multiple functions (Fig 8.3).

Fig. 8.2, Runway ML's Styles, 2024

Fig. 8.3, Runway ML's Functions, 2024

The Negatives:

Limited: The motion is very limited to a few movements: horizontal, vertical, pan, title, roll and zoom. (Fig. 8.4)

Aesthetics: uploading a video and changing its style looks clunky, and I could not use it for clients at this point. (Claymation is better done with clay).

Short renders: Generated videos are only 4 sec. long. They can be extended for another 4 seconds, but this is quite laborious and takes a lot of time to render.

Quality: The quality looks unsharp and cannot be used for commercial projects unless additional tools like Topaz (Chapter 10) are implemented to improve the quality.

Fig. 8.4, Runway ML's video camera movements, 2024

Fig. 8.5, Runway ML's image quality is unsharp, 2024

For video creation, there are various styles to choose from (Fig. 8.6):

Fig. 8.6, Runway ML's video styles, 2024

Overall, RunwayMl is one of the best available AI video tools right now, but it might soon get competition from Google's Video tool Walt, Sora by OpenAI or other. The AI world never stands still. Let's now look at another video option: Kaiber.

9. Kaiber

Estimated user base: 10,000+

Price: packs of $15 (1000 credits), $65 (5000 credits) or $120 (10000 credits) are available

Capabilities: AI text to video, video to video, and image to video generation with audio capabilities

Website: https://kaiber.ai

Example: https://vimeo.com/910867967?share=copy

Handbook: https://kaiber.ai/product

Kaiber is suitable for art and music video creation. Kaiber's mission: "to empower people everywhere to discover the artist within. We build generative creative tools and explore what it means for humans and machines to make art together. Our AI Research Lab and Creative Studio collaborates with artists and brands to develop immersive interactive experiences and tell stories in a whole new way."(Kaiber, 2024). I find Kaiber hard to control; therefore, it is limited to artsy projects that may not have a lot of specific requirements.

Fig. 9.1, The Author Kaiber video, 2024

You can choose one of three settings: Flipbook, Motion, or Transform (Fig. 9.2).

Fig. 9.2, Kaiber.ai, 2024

Flipbook: you upload a song and an initial image and Kaiber takes it from here. The results are a bit flickering like a rough cel animation and not smooth (Fig.9.1 and 9.4).

Motion: is supposed to be smoother than flipbook "Captivating fluidity in your animation"

Transform: is used when you want to transform an existing video's style (for example: live action) to another style (for example, 3d animation or claymation). The video length can be up to 4 minutes in the aspect ratio of the uploaded video.

The Positives:

Artistic: From my experience with Kaiber, it is pretty creative and adds elements that were not prompted. This is creative but also less controllable.

Community: Kaiber offers a gallery where you can showcase your work.

Pricing: Kaiber has flexible pricing and offers packs of $15 or $65. You pay as you need.

Audio reactivity: is a powerful tool that enables musicians and artists to enhance their music videos by synchronizing visuals with the rhythm and energy of the music. This can take your music videos to new creative heights.

Specific Styles: Kaiber is good at haunted/ scary images or manga-style illustrations (Fig. 9.3).

Fig. 9.3, Kaiber, 2024

On all devices: It is available as an app and can be used on Discord or a browser.

Longer videos: The amount of seconds that can be created is longer than other AI video tools.

User-Friendly: Kaiber is a user-friendly software.

The Negatives:

Limited: It is hard to control the outcome. It is creative, but maybe not in the style that you like.

Flickering: it has a flickering look and feel, even in the newer motion option.

Consistency: The style and visual consistency is clashing between keyframes.

Slow: It could use a bit faster processing time.

How to use Kaiber:

1. Go to https://kaiber.ai/create and select "Motion" as your video type.
2. Select either the 'Just start writing prompt' button or upload an image to start.
3. Prompt for the subject and style or use the preset choices on the right side.
4. Proceed to 'Video Settings' and set your video length (up to 16 seconds), choose your aspect ratio, and evolve (motion strength) parameters.
5. Click "Generate preview," and select one (1) of the four (4) preview frames presented—this will be used as a compositional anchor for the video.
6. At this step – if you are using pure text-to-video Motion – you can add additional Scenes (10 max) to extend your video up to 160 seconds. Note: this option is not currently available with image-to-video Motion.
7. Click "Create Video" and wait for your animation to complete.
8. Check the progress or access your videos by clicking "My Videos" in the navigation bar.

Fig. 9.4, The Author using Kaiber for a classical music video, 2024

Some of these already mentioned tools, are generating images in low resolution or unsharp. Here is a tool that can fix these problems: Topaz.

10. Topaz

Estimated user base: NA

Price Photo AI: $199 one-time fee

Price Video AI: $299 one-time fee

Website: https://www.topazlabs.com/

The gold standard in photo and video enhancement software, is designed for professionals and optimized for your local workstation. Topaz Labs' Gigapixel upscaling doesn't just look at neighboring pixels; it looks at whole sections of images at a time. (Topazlabs, 2024).

Topaz GigaPixel, Topaz Denoise and Topaz Sharpen are excellent tools for professionals to improve the quality of the image and upscale (Fig.10.1). However, they use much computing power and time to process (my computer gets hot when I use them), but the results are worth it. Using tools like this is not particularly environmentally friendly, and this needs to be taken into consideration (see ethics).

Topaz’ Photo AI

Photo AI tools are downloadable (not cloud based): DeNoise AI, Gigapixel AI, and Sharpen AI applications. Photo AI simplifies and enhances the photo editing process. The tools automatically examine the photos, recognize the main subjects and any quality issues, and then intelligently adjust the images with minimal effort required from the user.

Gigapixel AI 5.1 and later will also function as a Photoshop plugin. I have not tried this, as I took the easy route and dragged my image to the icon, and it scales up almost automatically. DeNoise AI is a noise reduction software that preserves fine details typically lost in traditional tools. Sharpen AI is a versatile tool that effectively addresses motion blur, missed focus, and out-of-focus images. It also offers minor noise reduction capabilities similar to Lightroom or Photoshop. However, it is important to note that this noise reduction feature is not as advanced as the AI-based, detail-preserving noise reduction in DeNoise AI (Topaz, 2024). It may be beneficial to have all Topaz tools that work in conjunction.

Key features

• Works with both Raw images and pre-rendered JPEGs
• Outputs DNG Raws for continued processing or ready-to-use JPEGs, TIFFs or PNGs
• Functions as a standalone app or a plugin for Photoshop and other apps
• Locates your subjects and identifies image quality defects automatically
• Corrects noise, blurring and lens distortions
• Upsamples low-res images to higher resolutions using AI algorithms
• Locates faces in your images and improves their visual quality
• Autopilot tool leaves little work beyond tweaking masks and adjusting strength sliders
• No subscription necessary
• Available for both Windows and macOS (dpreview, 2024)

Fig. 10.1, Image and video enhancing with Topaz, 2024

Gigapixel 7 is the newest Topaz tool. It sharpens, denoises, upscale and does light retouches:

Fig. 10.2, Image Enhancement with Topaz, 2024

Topaz’ Video AI

Topaz Video AI focuses solely on completing a few video enhancement tasks really well: de-interlacing, upscaling, and motion interpolation (Fig. 10.3). It's taken five years to craft AI models robust enough for natural results on real-world footage (Topazlabs, 2024).

Fig. 10.3, Video Enhancement with Topaz, 2024

The Positives:

The pros of Topaz are the features:

1. Upscale and enhance: This is probably the most useful feature for most video creators. It was previously not possible to upscale video without losing quality.
2. Motion Interpolation: is the process of synthesizing in-between images from a given set of images to create more fluid movement or to slow down the speed of a video file.
3. Stabilization: Can also be quite handy, when the camera was shaking when it recorded. Stabilization works by moving and rotating the original frames to make subjects visibly stable from frame to frame.
4. Deinterlacing: Deinterlacing is converting interlaced video into a non-interlaced or progressive form. Interlaced video signals are commonly found in analog television, digital television (HDTV) when in the 1080i format, some DVD titles, and a smaller number of Blu-ray discs. So if you want to improve old TV footage, this would do the trick.
5. Clean user interface
6. Format Options: Final images can be output as JPEGs, TIFFs, PNGs or DNGs, or you can render a partially-processed Raw file for continued work in another app like Photoshop.

The Negatives:

The downside of these tools is that they use a lot of power. My computer frequently heats up.

Other cons:

1. Compatibility issues: Not all AI video tools work seamlessly with every video format or editing software. Check compatibility before investing.
2. Limited video types: Some tools might struggle with specific video types, like fast-paced action scenes or complex textures. This can lead to artifacts or distortions in the output.
3. Unnatural results: While AI can enhance videos, it can also introduce artificial-looking elements, especially with features like face recovery or upscaling. This may not look natural.
4. Limited control: Unlike traditional editing, AI tools often offer less granular control over specific aspects of the video. You might have to accept the AI's interpretation, even if it's not exactly what you envisioned.
5. Artifacting and flickering: In some cases, AI processing can introduce unwanted artifacts or flickering effects in the video, reducing its quality.
6. Cost: Topaz tools can be expensive compared to traditional editing software. Consider the cost-benefit before investing.
7. Ethics and bias: Like any AI tool, potential biases in the training data can lead to biased outputs in your videos. Be mindful of this and use the tools responsibly.

Topaz offers very effective software with user-friendly interfaces. Plugins for Lightroom and Photoshop make its integration into any workflow easier.

Let’s investigate some more interesting AI tools that can make your life easier!

11. More Tools Worth Checking-out

There are hundreds of tools offering very similar features, and the market is overwhelming. I recommend using updated Tool lists, like Futurepedia or AI tools Directory (Fig.11.1).

Fig. 11.1, Video tools recommended in The AI Tool Directory, 2024

Here are a few of my favorite generative AI tools with specific capacities worth checking out for Text Creation, Image Creation, Branding, Advertisement, Video Creation, Productivity and Organization, Presentations, Audio, and Music.

Text Creation:

1. Perplexity is a great tool for research, because sources are given with each prompt result.
2. Claude: Claude is considered better than ChatGPT at consistently producing safe responses, because it is built with Constitutional AI. It is given a set of values that the answers have to adhere to and can then fine-tune itself using the constitution, rather than waiting for human feedback and input, which is how ChatGPT learns what responses it should refuse to answer (tech.co, 2024). Claude is trained to have natural, text-based conversations and excels in summarising, editing, Q&A, decision-making, code-writing, and more. I have been unable to test Claude because it is unavailable in all regions, but I recommend trying it if it is available to you.
3. Deepl: I tested DeepL for English-to-German translation and had more nuanced results than Google Translate or ChatGPT. I recommend checking it out for any language translations.
4. Jasper (formerly Jarvis): This versatile AI writing assistant generates various content formats, from headlines and ad copy to social media posts and scripts. It helps agencies overcome writer’s block, personalize content, and optimize for search engines.
5. Copy.ai: Similar to Jasper, Copy.ai focuses on content creation, offering templates and prompts for a wide range of marketing materials. Its multilingual capabilities cater to agencies with international clients.
6. Grammarly: This is a must-have tool for any writer. "Grammarly’s product ensures everything you type not only upholds accurate spelling, punctuation, and grammar, but is also clear, compelling, and easy to read." (Grammarly, 2024). This has been true in my experience using the free plan.

Image Creation:

1. Microsoft Bing: Bing offers a free option when you sign up with a Microsoft Bing account. It generates less biased images compared to other AI tools, as discussed in ethics.
2. Ideogram: It was initially supposed to come with improved type capabilities but still lacks perfect type generation (Fig. 11.2–11.4). The illustration styles are improving on this platform with more elaborate prompts (Fig. 11.5).
3. Leonardo: https://app.leonardo.ai/ I achieved good results with Leonardo, an image/video creator (free with a limit of 150 daily tokens) based on Stable Diffusion technology.
4. Photoleap: The phone app Photoleap generates surprising creative, but uncontrollable results. This can be useful when a client does not have any style preferences and you want to experiment freely.

Fig. 11.2, Ideogram image and prompt, 2024

Fig. 11.3, Ideogram prompt is not generating type properly, 2024

Fig. 11.4, Ideogram prompt: "Viva design Logo" has problems with letters, 2024

Fig. 11.5, Ideogram using an extensive prompt, 2024

Branding:

Canva: https://www.canva.com. It is an overall tool for beginners (not strictly an AI tool). Canva started as a digital template tool for non-designers, and it is easy to make quick prototypes with many template options. It is not a pure AI Tool. Great for presentations, productivity tools like whiteboards, websites, brand design, print products, videos, and more. It costs $12.99 per month or $119.99 when billed yearly. It now includes video tools Studio D–ID.

Leia: https://heyleia.com I have created a few website prototypes in Leia for quick ideation with clients. For that single purpose, it is perfect!

Fontjoy: Fontjoy.com It is useful to find font pairings.

Looka looka.com and Brandmark Brand.io are useful for initial ideation and prototyping; results cannot be used as final results as quality is still lacking. These are purely ideation tools with the promise that, in the future, you can have it all in one.

Huemint: Huemint.com inspires me with color palettes. This comes in handy when developing products from scratch. However, it gives the options of using the features Transformer AI, Diffusion Model AI, and Random AI without explaining what they are.

Designs.ai: https://designs.ai/ also offers a smorgasbord of tools to accommodate designers. The best tool here is face swap, where you can use the face of one model and place it over other models in different settings. This is not perfectly working yet, but I can see in the foreseeable future that photo models will be replaced by AI models for catalog shoots. This works only for very generic images, not for sophisticated, artsy photoshoots.

Mokker: https://mokker.ai/ Is good for creating mock-ups for products. It lets you easily replace backgrounds and looks promising.

Several tools promise to automate boring, repetitive design tasks. Tools change constantly, so it is important to go through trial and error to find the right ones for your own purposes.

Advertisement

Dragonfly: https://dragonflyai.co/ This platform could be useful for testing your creative ads, e-commerce optimization, and analyzing data-driven insights.

Adcreative: www.adcreative.ai/ is for optimizing Ad Performances and generating ads and banners that bring conversions.

EyeQuant: is an advanced, professional AI-powered platform analyzing visual attention on ads and creative materials. Predicts eye movements and areas of interest, helping optimize visual elements for stronger impact.

Video Creation:

Pica: Some designers prefer Pika Labs over Runway for generating videos with AI. The process for animating images involves selecting the motion brush option and painting the part of the image to be animated. Pica Labs is another good text-to-video generator, while Runway is better for animating images. Pica, like MidJourney, lives on Discord, which leads to potential inspiration and idea theft. The prompt is: /create

Studio d-id: https://studio.d-id.com/ I use this platform for presenter-style talking head creations. You can turn photos or AI portraits into lifelike, talking avatars and generate presenter-led videos.

Gooey: https://gooey.ai/explore/ offers customizable QR codes, animations, and more (Fig. 11.6)

Synthesia: Synthesia.io is a tool to create narrator videos quickly. It generates the script, an AI voice, AI avatars, and a video editor. The AI voices sound artificial, and the look is semi-professional, but over time, this may improve and can be used for explainer videos.

Pictory can turn an article into a youtube video.

Luma: https://lumalabs.ai/ promises to create 3d animations and models, but my simple prompt "Create a spinning discoball" Did not yet achieve an amazing result. I would keep an eye on the app, since it is developing fast.

Veed, HeyGen, Capcut and Steve.ai are other video tools that you can try out. In the near future, Google might release Walt (Warping Artificial Intelligence for lifelike transitions), which operates a bit differently by crafting a 3D model based on a static image. This might be cool to check out once it is released.

Fig. 11.6, Gooey's features, 2024

Productivity and Organization

Taskade: https://www.taskade.com, as mentioned in the case study, is good for organizing teams

Adobe Sensei: Integrated within Adobe Creative Cloud, Sensei leverages AI for intelligent workflows and automation. It analyzes assets, recommends edits, and generates variations, enhancing design efficiency and consistency.

Notion: While not strictly an AI tool, Notion empowers agency teams with collaborative workspace, project management features, and AI-powered search and suggestions, fostering organization and information flow. I used Notion for this research book, and it saved me much time, because I did not need to scroll as with word doc, indesign or pages.

Audio Pen: is a tool where you can dictate with your voice and it will create a professional sounding text (Fig. 11.7)

Fig. 11.7, Audio Pen, 2024

Presentations

Beautiful.ai: has a user-friendly interface, and focuses on aesthetics and design automation. it is less flexible for complex presentations, and data visualization features limited.

Tome: you input a prompt to start presentation creation. It enhances AI-powered content. But it has fewer templates and design customization options compared to others.

Simplified: has robust AI features, and generates presentations, images, and more across various mediums, but comes at a higher price point compared to some options.

Gamma: has strong data visualization capabilities, and it integrates with business intelligence tools. Lacks some creative flair and storytelling elements found in other options.

Audio and Music

Mubert: https://mubert.com/render for creating copyright-free music. This platform generates original music based on various styles and moods. It’s perfect for composers and creators looking for unique instrumental pieces for films, podcasts, or games.

Murf: This text-to-speech platform offers various AI voices for creating voiceovers, podcasts, and audiobooks. It’s ideal for content creators, marketers, and educators seeking high-quality narration.

Landr: https://www.landr.com/ This advanced AI tool analyzes and masters your music, enhancing audio quality for professional results. It caters to musicians, producers, and audio engineers seeking a polished sound.

Otter: I have used Otter for a long time, which is useful for transcribing audio in meetings or videos.

LalaAI: https://www.lalal.ai/ This tool separates vocals and instruments from audio files, helping musicians isolate tracks for remixing, sampling, or creating mashups.

Fig. 11.8, Ideogram image and prompt, 2024

Conclusion

I recommend developing your style with and without AI, embracing different tools, and mix & match techniques. Craft your legacy, not copy-paste. Sense and adapt to ever-shifting trends. Your human intuition & empathy are superpowers – leverage them to stand out in the creative landscape. With AI tools, you will have an advantage over traditional designers. AI tools boost creativity and make you exponentially more productive.

Canva or Adobe Express offer cookie-cutter solutions and are valuable assets for low-budget solutions. However, specialized tools will achieve more professional results, and training your own data sets will give you an edge.

Even though most AI tools are user-friendly and easy to work with, it does not mean it will not take time to get worthwhile results. There is a learning curve closely connected to the time you spend experimenting.

MJ will likely become market-dominant with worthwhile results for designers. Photoshop AI, Adobe Firefly, Stable Diffusion, Dalle-3, and Leonardo are other top image-generating AI tools. The landscape of AI apps is rapidly growing, with more apps constantly being developed to empower designers.

Consumers also should know more about AI image creation to distinguish reality from fiction, real and fake news. AI developers should watermark AI content. The legal landscape surrounding copyrighting AI-generated images is currently quite complex and murky, with no definitive answer. I recommend to use AI primarily as a prototype tool, for ideas and not for the final exposure, read recommendations for more details.

Go to the next chapter for a visual demonstration of generative image and video results!

12. Video demo

I created numerous AI-generated images and videos, to document the state-of-the-art as of the date of publishing, demonstrated in this video montage (Fig.12.1):

Fig. 12.1, The Author’s Tool Exploration, 2023

In addition to showcasing what AI tools are capable of, I also wanted to give you examples to get started.

13. Prompt examples

In which areas of work are the most successful image generations? In this chapter, I showcase a few prompt examples to give you ideas in the areas of Concept Art, Future Scenarios, Photo-Realistic Images, Illustrations in Manga or other styles, and Mock-ups. As you can see, a lot of prompts are very specific, with descriptive adjectives and references. This is important to keep in mind when prompting!

Great work can be achieved in the category of concept art, a visual art form used to convey ideas for films, video games, animation, comic books, television shows, or other media. AI tools are very effective in creating looks, styles, and designs before starting production for the final product. Concept art usually refers to world-building artwork, which makes sense to do with AI.

Concept Art: Monsters, Aliens, Characters

MJ Prompt: Wes Anderson style, photo of anthropomorphic dog dressed as a businessman going to work on Monday morning --ar 3:4 -- stylize 1000 --v 6

MJ Prompt: macro close-up of a cosmic sea slug in a dark sea with eyes and a grin --ar 9:16 --v 6

Prompt: Wes Anderson style, photo of anthropomorphic dog dressed as a businessman going to work on Monday morning --ar 3:4 --stylize 1000 --v 6

Prompt: the face of an alien, with metal bands around his eyes, in the style of machine aesthetics, sculptural pictorialism, gothcore, futuristic robots, michael hussar, dallmeyer 3b 75mm f/1.9 super six anastigmat, dan hillier --ar 51:64 --v 6

Prompt: chicken with hat wall art reversible canvas print, in the style of grisaille, wallpaper, rosa bonheur, pinturicchio, toby fox, hand-painted details, quito school --ar 5:7 --stylize 750 --v 6

Prompt: Morden art, View from mid front, canon EOS R10, 135mm lens, helmet split in two half, left half human bronze skeleton, right half tall kaws human like alien compound eyes , half leather, half jumpsuit , louisvuitton design in 80s, rich Pastel vibrant colours, sharp Rich colours on display in a white art gallery, studio lights, sharp light, bright, Georges Antoni photography, epic , unreal engine 6, high end finish, photo realistic, uhd, 8k --ar 2:3 --v 6

hubby cute fluffy little Pomeranian's belly, funny facial expressions, exaggerated movements, 3D characters, white background, a bit fluffy, elongated shapes, cartoon style, minimalist – chaos 12 – iw2 – AR 3:4 --ar 3:4 --stylize 250 --v 6

Concept Art – Future Scenarios

MJ Prompt: https://s.mj.run/wBOj_c9gm7U Richard Meier, 2-building, --ar 4:3 -- stylize 250 --iw 2 --v 6

MJ Prompt: a green structure with a bridge on it in the middle, in the style of transparent layers, uhd image, emily carr, liam wong, piles/stacks, mountainous vistas, stop-motion animation --ar 53:64 --style raw

Photo-Realistic

Prompt: Modern Office space with Vertical Garden Wall and large windows --ar 16:9 --v 6

Prompt: https://s.mj.run/dD9YaXA5iMU https://s.mj.run/k5K3fxzT07U Extremely realistic luxury bedroom photos, 3.3m high, super minimal, modern interior, bright atmosphere, realistic material, king-sized bed, cool sensation, studio type floor plan, middle shot, focus on side table, trending on air, --ar 9:16 --v 6

MJ Prompt: Studio photo of a white expedition tent in the form of a dome on a white background, strictly frontal view, pulsed studio lighting, background uniformly white, ground level view, full – body shot, --ar 16:9 --style raw --niji 6

MJ Prompt: a sketched diagram depicting "Different types of Neurons in action", dry brush, white on black background, --ar 16:9 --style raw --stylize 150 --v 6

MJ Prompt: whale shark swimming in polluted ocean surrounded by plastic bottle and trash photography --ar 16:9 --v 6

DALL·E PROMPT: A creative and intricate diorama depicting a miniature cityscape built upon an Adidas sneaker. The sneaker serves as the base of the diorama, with minimal background

MJ Prompt: coloring space, alien aircraft carrier, Miami in the background, futuristic fighter jets, simple, clean lines, black and white --ar 4:5 --niji 6

MJ Prompt (by michbat): a woman with a Siamese fighting fish covering her eye, in the style of rendered in cinema4d, ethereal nature scenes, lifelike figures, flora borsi, olivier valsecchi, thomas benjamin kennington --ar 3:4 -- style raw --stylize 250 --v 6

Product Mock-ups

MJ Prompt: a woman's hand is holding a white mockup tube of facial cream on a pink isolated background --ar 16:9 --v 6

MJ Prompt: coffeecup and white paper on wood desk, white paper mockup --v 6

14. New workflows with AI tools

Designers using AI tools may consider creating their own unique workflows combining various tools: the more individual the workflow, the more unusual the results. If only one tool is used, results may look predictable, generic, and flat. The more you experiment with tools and train your own data, the more distinct and creative the styles become; see samples below (Fig. 14.1–14.3). I want to encourage you to explore freely and not get too attached to the outcome but to view the work with AI as an iterative process to get better results over time. These are, after all, only tools and ideas start in your brain.

AI can be utilized to push creative boundaries, not just to automate processes and save time. The creation process can actually become time-consuming.

Fig. 14.1, Mitra Allahmoradi Instagram, 2024

“It takes a village to put together one image, the setting, lighting, mood and attitude, design ethos, the humor, the skin. AI is a new layer of communication that is one big impactful tool.”

Ceren Arslan, 2024

Fig. 14.2, Ceren Arslan, Architectural interiors, 2024

Fig. 14.3, Gemmy Galeno, Midjourney, 2024

Example workflow:

I used ChatGPT to improve my narrative and script for a talking head video. I then created an avatar with MidJourney. The avatar image and the final script were imported into studio D–ID to make a talking head. Since the AI voice provided in D–ID sounds artificial, I trained my own AI voice on RunwayML and then uploaded that into Studio ID to combine the avatar with my voice. The result (Fig. 14.4):

Fig. 14.4, The Author’s workflow with Talking heads, 2023

I also trained two movie star voices on RunwayML to create a clip about Barbie and Ken in the Barbie movie as a test. The faces are talking and moving, but the bodies are still, which gives this an unnatural, eerie look. Is cloning voices for entertainment ethical? I feel very torn about this, since it can lead to scams and damage reputations. This 2024 example shows AI limitations, how technology has advanced, and even more so, where things are going. In this rapid AI development, a year from now, we may assume that bodies and movements will look more natural. Here is the somewhat questionable result (Fig.14.5):

Fig. 14.5, The Author’s workflow with Talking heads “Barbie”, 2023

Working with clients and setting expectations

When using AI tools, it is crucial to learn how the AI systems work, their strengths and limitations to avoid false expectations. A violinist came to me asking for an AI-generated image with his photo as the basis of the prompt. This request exceeded the capabilities of MidJourney and other tools. AI tools have difficulties with hands and fingers, arms and legs. Holding a complicated instrument in a specific way was simply too much to ask (Fig. 14.6).

Also, the client did not have style references. It is important to learn the language of prompting with references and not assume AI will figure it out by itself. MidJourney can not mindread, therefore, it is impossible to expect a brilliant result when the request is not specified. When working with a client it is essential to set expectations and do some training beforehand.

Fig. 14.6, Difficulties with hands, positions, details in MidJourney’s images, 2024

A prompt should include:

• Subject: Clearly describe what you want to see in the image (person, animal, object, scene).
• Style: Specify the desired artistic style (painting, illustration, photorealistic, surreal, etc.).
• Details: Be specific about any important details you want included (color palette, mood, composition).
• Light conditions: at dawn, at dusk, bright daylight, etc.
• References (optional): Include image URLs or specific artists/artwork if you want similar style or elements. If you are aiming for a photorealistic look, you can add the desired camera, lens (fisheye, wide angle, etc.)
• Foreground, background info

Clarity and Focus:

• Use concise language: Avoid overly long or complicated phrases.
• Prioritize key elements: Highlight the most important things you want to see.
• Be consistent: Avoid contradictions in your descriptions.

Additional Tips:

• Experiment with different wording: Try different descriptions to see how they influence the output.
• Start simple: Begin with basic prompts and gradually add complexity.
• For multiprompts use :: (two double dots) in between phrases (MidJourney).
• Use parameters: Utilize features like -ar for aspect ratio or -q for image quality. See more in the handbooks of the AI tool that you are using.
• Explore the community: Learn from other users’ prompts and examples.

Here are some examples of different prompt types:

• Simple: “A cyberpunk city at night, neon lights reflecting in rain-slick streets.”
• Detailed: “A portrait of a woman with flowing red hair, wearing a golden crown, in a style reminiscent of Klimt’s paintings.”
• With reference: “A painted image of a cat playing with a ball of yarn, similar to the style of Andy Warhol’s pop art.”

Remember, the key is to be clear and specific, and experiment to find the prompts that generate the images you or your clients desire! Do not get frustrated when things do not work out as AI is in development and quickly improves. To find out how to use AI in prototyping, read on.

15. How to prototype with AI tools

To investigate AI tools in practical contexts, I use a simple process often used in agile workflows (Fig. 15.1): design, test, detect faults, improve design, test the altered prototype, and so forth. I am not going into details about Design Thinking or Agile, but many brand designers use these methodologies, which work exceptionally well with AI tools.

Fig. 15.1, Justinmind, Iterative process, 2023

AI-generative tools can enhance creativity throughout the design thinking and prototyping process in several ways.

Here are a few examples:

1. Ideation and problem framing: AI can assist in the ideation phase by generating a wide range of creative solutions to design challenges. By analyzing user data, market trends, and design principles, AI generates ideas that human designers may not consider. This can help to break out of your comfort zone and explore new possibilities.
2. Prototyping and exploration: AI can accelerate the prototyping process by automatically generating sketches, mockups, and prototypes based on design specifications. This can free designers to focus on more creative aspects of the design process, such as user research and interaction design. AI can also help designers explore a wider range of design options by generating a multitude of variations.
3. Rapid iteration and experimentation: AI can facilitate rapid iteration and experimentation by quickly generating new design solutions based on feedback and test results. This can help designers quickly identify and refine their designs, leading to more polished, effective solutions.
4. Personalization and adaptation: AI can personalize design solutions to meet the specific needs of individual users or groups. This can be done by analyzing user data to identify patterns and preferences, and then generating designs tailored to those preferences.

Fig. 15.3, Design Thinking Model Stanford School, 2024

How can AI tools be used to enhance creativity in design thinking? (using the Stanford School Model, Fig.15.3)

1. Empathize phase: AI can help designers understand user needs and pain points by analyzing social media data, customer reviews, and other sources of user feedback. In the first phases of the design process, text-to-text tools help empathize, create customer personas, find ideas, write the creative brief, list deliverables, and more (Fig. 6.20). ChatGPT offers an avalanche of design ideas that a single designer could not develop as quickly. This represents a significant jump-start to the creative process. In addition, AI production tools such as Taskade assist with the organizational framework.
2. Define phase: AI can help designers frame design challenges by analyzing market trends, competitor offerings, and user data.
3. Ideate phase: Designers can generate a wide range of creative solutions by analyzing past projects, design principles, and user data.
4. Prototype phase: AI can help designers create rapid prototypes of their designs by automatically generating sketches, mockups, 3D models, high-fidelity models, graphics, or simulations. AI-generated visuals can be easily modified, saving time and resources compared to traditional methods. While logo designs may not be final, design elements or photographic images are created. AI tools are excellent for finding styles but not for completing a brand design. Designers must still rework AI results with digital or traditional tools for a consistent, scalable brand design. MJ excels at creating stylized, photorealistic images but has limitations with numbers, letters, and creating details. Quality is about details. Color palettes and type options can be effortlessly created with AI, inspiring the designer’s imagination. Most commercial design projects require “mediocre work” that can be done quickly and effortlessly with AI (Fardost, 2023). I am not sure about producing mediocre work, but making money with design may sometimes force designers to push work out the door. For brilliant work, designers can further push their creativity with AI tools.
5. Test phase: AI can help designers gather feedback on their designs by automatically generating surveys, A/B tests, and user interviews. AI quickly creates quantity, which is important for “seeing” and inspiration, as mentioned earlier. It rapidly creates design variations that can be tested in focus groups or AB testing. AI tools like Dragonfly help analyze and test, significantly reducing cost, great for design agencies that must balance creativity and budgets (Fig. 15.2).

Fig. 15.2, Dragonfly AI platform, 2023

Other benefits of using AI tools in Agile & Design Thinking:

• Cross-Disciplinary Inspiration: AI can draw on a vast array of sources and disciplines to inspire designs, breaking the silos of traditional design thinking. For example, an AI could suggest design elements based on nature, art, science, or other industries. It can blend images like nature and architecture and lead to innovative designs.
• Accessibility and Inclusivity: AI tools can make design more accessible to non-designers, encouraging a more inclusive and collaborative approach to problem-solving, which aligns well with the collaborative ethos of design thinking.
• Customization and Personalization: AI can tailor designs to individual users’ needs or preferences at scale, something incredibly resource-intensive without technology.
• Sustainability and Efficiency: AI can optimize designs for sustainability, analyzing materials, and processes for environmental impact, which is increasingly important in modern design processes.
• Predictive Analysis and Future Forecasting: AI can predict future trends and user needs, helping designers stay ahead of the curve and create forward-thinking solutions.
• Error Reduction and Quality Improvement: AI can identify potential flaws or areas of improvement in designs, enhancing the overall quality and effectiveness of the end product.

By integrating AI with design thinking, the creativity and effectiveness of the design and prototyping processes can be significantly enhanced, leading to more innovative, user-centered, and sustainable solutions.

Let us look at the case study to see if the tools work in a professional context. I do not want to spoil it, but the outcomes were very mixed.

16. Case Study – The Real Hero

Can AI tools generate a creative, usable branding package for a client? Can you, as a designer, make money with AI-generated designs? To find out, I worked with a Swedish start-up called The Real Hero (TRH), which approached me in August 2023 with a list of desired brand deliverables (Fig. 16.1). We agreed to try AI tools to prototype the deliverables. The expectations were set to produce prototypes, not final designs. TRH wanted to find out if they should use AI tools for their clients. Since 2023, many things changed and some of the tools might be optimized by now. This is testament to how fast development is moving in the AI space.

Fig. 16.1, TRH, Brand Deliverables, 2023

The TRB Brief is an excellent starting point for the branding effort.

However, for a more comprehensive branding list, I asked Taskade AI for guidelines. It generated a useful, complete list with check-boxes for timelines, due-dates, team assignments, all productivity tools for project management (Fig. 16.2).

Fig. 16.2, The Author, Taskade, 2023

Logo Development

Designs.ai

I used designs.ai for logo development. After submitting name, tagline, icons, and desired color palette, it generated several logos in a few seconds (Fig. 16.3).

Results: The designs are simple, cut-and-paste, and not usable. I tried another AI tool.

Fig. 16.3, The Author, designs.ai, 2023

Logo.ai

Logo.ai asked to follow these process steps, which I did (Fig. 16.4):

Fig. 16.4, Logo.ai, 2023

Result: Generated logos were not in the chosen color palette; they omitted or confused the tagline, and are not inspiring.

MidJourney

I moved the exploration to MJ (Fig.16.5).

Prompt: “A black and white logo vector graphic for THE REAL HERO”

Result: Four portraits of a bearded man with a retro logo with illegible letters. Apparently, the algorithm sees a real hero as an outdoorsy man with a beard in retro style.

Fig. 16.5, MJ, 2023

Leonardo.ai

Prompt: a logo for a branding company called “The Real Hero.”

Result: It generated a male, Marvel-style superhero in American flag colors (Fig. 16.6). “Hero” is interpreted as male even though “hero” is officially gender-neutral. To test this further, I prompted: “Logo for a branding company called The Real Heroine,” which showed young women, only white, no other races (Fig. 16.7).

Fig. 16.6, Leonardo.ai, 2023

Fig. 16.7, Leonardo.ai, 2023

Gender, age, and race bias are apparent in the algorithm. The type has not been generated correctly. What if I omit the word “hero”?

Prompt 3: “logo for a storytelling branding company”

Result 3: has a retro, European/Anglo-American flavor and does not design a diversity of forms, shapes, or colors, nor does it include different cultural aesthetics like Asian or African designs (Fig. 16.8)

Fig. 16.8, Leonardo.ai, 2023

Brand.io

I tried one last time to create a usable logo with Brand.io. The process is similar to designs.ai and logo.ai (Fig. 16.9).

Fig. 16.9, Brand.ai, 2023

Results: The generated logos lack personality but are slightly more usable than previous results. The chosen color palette was not taken into account. To download the logo, I would need to pay $65.

Color Palette

AI color generator huemint.com is designed to make color choices with mock-up options to see how colors look on merchandise. I aim for minimal colors, per the creative brief (Fig. 6.10).

Fig. 16.10, Huemint, 2023

Results:

1. Black, grey, pink
2. Green brown, yellow
3. Red, grey black

Conclusion: Huemint has a narrow purpose, which is why it works well. It does what it promises and helps with color choices.

Fonts

Like Huemint, Fontjoy has a narrow purpose: find a font pairing (Fig. 6.12). A randomizer suggests Google font combos (Fig. 16.11).

Conclusion: Fontjoy works well for finding the right type combo, improving the design process.

Fig. 16.11, The Author, fontjoy, 2023

Fig. 6.12, Fontjoy’s Typevisualizer, 2023

Graphic Elements

Since the TRH Brief asks for simple one-line drawings, I prompted the coffee cup with the Picasso image from the clients’ brief (Fig. 16.13).

Fig. 16.13, Picasso’s sketch from brief, 2023

Prompt 1 result: vector clipart without much character, Fig. 16.14.

Prompt 2 results: Prompt 2 added the word Picasso (Fig. 6.16) and the results vaguely reminded of the artist. I “saw” something in Image 2, and asked for variations (Fig. 6.17).

Prompt 3 results: Variations, Fig. 6.17

The results became more interesting.

What if I want to create another element (camper-van) in the same style?

Using the seed-number, the result is stylistically close to the coffee cup. However, it is too detailed and without Picasso’s touch Fig. 6.18

With a simplified prompt, the result was achieved, however, without Picasso’s style, Fig. 6.19

Conclusion:

AI is useful for initial brand project explorations to determine style, but final designs require a designer’s touch for more inspired results, especially in vector formats.

Not all projects benefit from AI; some are better suited to digital or analog tools. For instance, the TRH design elements are more effective when hand-drawn, the AI results were mediocre.

I suggest a step-by-step approach to building branding elements using multiple narrow AI tools:

1. Type
2. Color palette
3. Logo
4. Graphic elements
5. Website, etc.

MJ excels in creating stylized, photorealistic images but struggles with numbers, letters, vector-style graphics, and logos. It offers various parameters and styles, but identifying the right prompts is time-consuming. Other AI platforms like Brand.io, logo.ai, and designs.ai promise complete branding packages but fall short in style and sophistication.

Generally, more expensive, high-end tools might deliver better results. AI tools designed for specific tasks are more effective.

How do TRH results score on the creativity test?

1. Fluency – 10
2. Originality – 3
3. Flexibility – 2
4. Elaboration – 9
5. Sparkle – 5

(1 low, 10 high)

Client Feedback

TRH’s feedback was an evaluation of the various AI design tools. Taskade was evaluated as potentially useful for TRH to organize creative tasks. Designs.ai was criticized for its basic output and unappealing color choices, akin to simple designs from Canva, although a couple of elements seemed promising. Logo.ai disappointed due to the absence of specific industry representation and generic, watermark-laden logos. MJ, however, was praised for producing high-quality, eye-catching images, despite not representing the company. The ‘American’ output from Leonardo.ai and the poor color consideration from Brand.io were also mentioned. Huemint.com was noted for practical color applications, and Fontjoy for its novel font pairing map, which was well-received for visualizing and comparing fonts.

TRH’s evaluation concluded with their skepticism about the efficiency of AI tools in a brand agency context, suggesting that the time to select and learn the tools outweighs their potential benefits. Text-to-text tools seemed superior to image-creation tools.

The feedback concludes with a preference for human-driven design solutions for the time being.

We saw in the case study that not everything worked out; in fact, only a few things were usable.

Just 6 months after this case study, I am again experimenting with logo designs with AI. This time, I had better experiences, which shows improvement in the AI-type capabilities. Here are some logo examples where I mixed the use of AI and Digital tools for my company Viva Design (Fig. 6.20).

Fig.6.20, The Author’s Logo explorations with AI, 2024

The mixed results in the case study led to the question: can you enhance your creativity with AI tools?

17. Can you enhance your creativity with AI tools?

Yes, you can.

You can generate stunning, detailed, and beautiful images with AI. AI can generate a large quantity quickly, boosting the design process. Even less exciting images can at least inspire designers and enhance their creativity. ChatGPT excels in all creativity tests and exponentially boosts creativity. It is simple to prototype with AI, and the perceived risk of failure is low. This invites self-proclaimed non-creatives and beginners to experiment and create more freely, jump-starting their creativity. AI’s user-friendliness invites you to expand your repertoire and try new skills like animation.

For design agencies or departments, AI tools effectively define customer needs, profiles, and demographics. They’re valuable for generating prototypes, mood boards, ideas, concepts, strategies, product transformations, services, and organizational development. In that sense, AI tools boost creativity and productivity.

However, here are some thoughts to consider:

1. AI is not a “magic bullet”. Rather, it is an imperfect tool that can be used in design, agile work processes, design thinking, and many applications and methodologies. It takes time to learn, and the process involves a lot of trial and error. Creating something random and “cool” with AI is easy, but controlling details and finalizing designs is challenging. I suggest using AI tools primarily for ideation and prototyping. Clients should be briefed before the project that results may be experimental and serve as prototypes, not as final outcomes. For a final brand design package, I would accelerate ideation with AI but rely on my own ideas and traditional and digital tools to finish.
2. Regarding text-to-image tools, prompting can have unpredictable results that seem like “magic” (Gilbert, 2016). A surprise or mistake can be a porthole to creativity, as discussed earlier, but what if it is not? Crafting a detailed prompt ensures a better, more controlled result.
3. Users become more aware of what they want during this prompting process. This adds additional value and clarifies the mission. Results can influence the quest. In a Viva Design logo project, the presented logos inspired self-reflection, and the client requested more logos to “find themselves.” This process taught me that designers can become “therapists” and help organizations “find themselves.” I presented over 50 logos over four weeks. This process was expensive and time-consuming. Using improved AI tools, it would take a day to create 100 logos and another day to finalize the design. Using AI is not only creative but also highly cost-effective.
4. Can designers use their AI-generated images for commercial products or services? “United States copyright law protects only works of human creation.” (Judge Howell, 2023). When publishing AI-generated images, there is no protection unless the images have been altered; creators must exert intellectual effort—another reason I suggest using AI images predominantly in prototyping.
5. Ethical considerations must be considered when using AI tools responsibly. See Ethics.

Regarding text-to-text tools, creativity excels. It is only a matter of time for text-to-image tools to catch up. I recommend organizing thoughts, ideas, and critical thinking before using ChatGPT and other tools to prevent losing track. As powerful as AI is, the human designer must focus empathetically on the client’s needs and keep human interests at the center. In the next chapter, I offer recommendations that draw from my practice as a designer and my research.

18. Recommendations

This chapter highlights recommendations for individual brand designers, creative agencies, and AI developers as a way to move forward with new AI knowledge and skills. Education, training, and workshops are recommended to stay up-to-date for everyone, if you are a designer, design manager, entrepreneur, or client.

Recommendations for Designers

To effectively integrate AI into the graphic design workflow, I recommend these steps:

• Identify repetitive tasks: Tasks that consume significant time but involve repetitive actions are ideal for AI automation.
• Research various tools: Evaluate their compatibility with your existing workflow.
• Conduct and/or attend workshops: Share your AI knowledge regularly and make it a creative “habit.” Learning from other’s trial and error saves time. For workshops, you may go to ulrikekerber.com/aiworkshops.
• Experiment and prototype: Start small by incorporating AI in specific areas of your design process. Evaluate the impact, refine, and iterate.
• Protect your art: Designers can protect their artwork to combat intellectual property theft via applications such as Glaze, a high-tech “cloaking” technique (Techcrunch, 2023).
• Be copyright-aware: Since the copyright laws are murky, I recommend using AI-generated visuals for prototyping, brainstorming, and internal use, where copyright concerns might be less stringent. Alter your works significantly, modify, or combine AI-generated images with your own creative input to create truly unique works with stronger copyright claims. Collaborate with other designers or artists to create collaborative works, potentially strengthening claims of joint authorship. When in doubt, seek legal counsel from intellectual property lawyers specializing in AI for specific guidance related to their projects and copyright claims.
• Secure your job: Expand your skill set, develop a unique aesthetic, and make yourself valuable (beyond skills and tools). Successful designers have taste and an opinion (Fardost, 2023) and are not afraid to show it. Expand your skills by experimenting with new tools and workflows.
• Stay informed: Rules and laws change. Stay updated and informed about ethics, and read the upcoming “AI handbook for designers”.

You should be cautious when relying on AI-generated images, as their origins may not always be disclosed. When used responsibly, AI can be creative, productive, and “do good.”

Fig. 18.1, Deseen, Nutella, 2017

Recommendations for team leaders in creative agencies

• Be productive: Working with AI increases creativity and productivity. Comparing one month’s worth of traditional image creation versus one hour of design with AI, it is easy to see how much more can be done at a considerably lower cost. AI’s strength lies in large quantities and speed. Design 100 sneaker images in a day or create a million random graphic patterns for packaging (Fig. 18.1)! Adobe Express and Canva use AI to design layouts, colors, and typefaces based on prompts. This saves time and effort, allowing designers to focus on more creative, complex tasks like aligning brand elements.
• Analyze: AI assists in analyzing market data and identifying patterns in user behavior, enabling them to make informed decisions about which problems to address (Berkovicz, 2023).
• Strategize: AI strategies should incorporate a roadmap with clearly defined time bounds, ROI targets, and a budget that includes AI education, such as ongoing AI workshops.
• Educate and share: Create opportunities for updating skills, onboarding newcomers, and sharing the latest news and ethics in workshops. Team members benefit from exchanging new AI developments and skills and connecting in a stressful, digital, lonely, AI-infused world. A clear workshop structure and a focus on a single tool and theme are best. Make it a series to strengthen the creative habit (Tharp, 2015), as the creative “muscle” needs a regular workout. Continuous training boosts creativity, productivity, and performance (Seismic, 2023).

The future of AI in brand design holds immense potential. Embracing AI will empower designers to streamline their workflows, unleash their creativity, elevate the quality of their work, and stay at the forefront of the industry.

Fig. 18.2, The Author, Creative Applications and Uses of AI, 2023

Recommendations for AI Developers

• AI disclosure: “If I’m having a conversation with someone and I cannot tell whether this is a human being or an AI, that’s the end of democracy because that’s the end of meaningful public conversations (Fig. 18.3, Harari, 2023). AI art should be visible as AI art! As a regulation, make it mandatory for AI to disclose that it is an AI. AI Tool providers should prioritize transparency.
• Bias: AI Tools should only be released after careful safety and bias checks. The algorithm’s bias needs to be eliminated. Potential risks and ethical issues must be addressed to ensure AI does not become more harmful to humans.

AI tools can help explore new possibilities, services, and products that help envision the future (Fig. 18.2). However, with AI in the long term, designers and agencies may lose some of their traditional work and face bigger issues, see next chapter.

Fig. 18.3, Video: The AI Dilemma, Raskin, Harris, 2023

Recommendations for Beginners or Design Students

AI tools can be overwhelming, and the options to create ideas are endless. This can be confusing if your mind is not clear. Before you start working with AI tools, you may want to think and write down:

• Spend 10.000 hours to become a real designer. Design is a set of skills, and mastering AI tools is just one aspect of many. Get skilled so that you can execute what you want to achieve. What skills do you need to improve on?
• Define your goals, both creative and general life goals, so that you will not be overrun by things that just happen to fall in your lab. Design your life. Create a mindmap.
• Find like-minded collaborators. Surround yourself with people who care about you. Who would you ideally like to work with, and how can you make that happen?
• Consider clients that reflect your vision. List ideal clients and projects.
• Who could be a good mentor for you, who challenges and nurtures you?
• Know your references. To write a client brief or a proposal or to communicate with a client or collaborator, know the styles you can refer to. This will become essential in AI prompt writing.
• Always stay curious and continue to learn. Ask questions, and do not shy away from new tools or environments. What tool or project makes you curious now?
• Try to be aware of your environment, the organization you are in, and yourself. Breathe.

Recommended Workshops for Continued Learning!

Online workshops and courses have risen in popularity in recent years. However, over 90% of learners drop out of open online courses, according to a study by Springer (Springer, 2023). Self-timed online courses seem too complicated, long, and lonely for participants to stay engaged.

I tested live, interactive, short-format online workshops with hundreds of participants designed for professional creatives and technology-reluctant participants working in the creative field.

Fig. 18.4, Amy C. Edmondson, 2023

The Benefits of Workshops

• Workshops can offer AI onboarding for any imaginative individual, not just designers, “democratizing” design. This could promote team diversity. Harvard professor Edmondson recommends diverse teams for solving problems (Fig. 18.4).
• Workshops welcome an inclusive, engaging, ethical discussion, in the midst of AI’s debated, negative perception.
• Workshops transmit the skills needed for the job market. AI is a much needed skill for the job market (Fig. 6.23). There should be great interest in learning AI.

Workshops should create a collaborative and productive environment to ideate, experiment, and prototype, balancing human ingenuity and the responsible use of technology. Knowing about overwhelm, reluctance, and fear, I design my workshops as easy-to-follow as possible. Participants should overcome barriers and learn practical tools that inspire further exploration.

Regular weekly short workshops (under 1 hour) are a great way to introduce and onboard AI-tools and ethics. Everything over an hour cannot be justified in a busy agency context. “Learn creative tools in the time it takes to eat a sandwich”, is my Motto.

Workshop invitations should sound welcoming and offer a clear take-away like “You will learn to create seamless patterns.” To make them fun, exercises like creating a “future transportation hybrid” with excellent examples should be prepared. Feedback is important for iterations. However, participants need special encouragement to volunteer feedback. Most participants were inspired and left “wanting more.” Short workshops inspire, but learning skills for the job market requires more immersion.

To article a workshop or watch an AI workshop video, please visit Ulrikekerber.com/aiworkshops (Fig.18.5).

Fig. 18.5, The Authors’ workshops, 2024

Let me finish now with some final thoughts.

19. Final Words

AI has transformed the world, unlocking numerous possibilities. It revolutionizes how we communicate, create, and work. It rapidly and profoundly impacts all sectors of society, from business development to medicine, education to research, and coding to the arts. Like many other transformative technologies, AI offers enormous potential but also carries significant risks.

By being a brainstorming partner, AI can help your team be significantly more creative and productive. Used in ideation and design sprints, it dramatically reduces time and cost. It triggers new ideas with relevant stimuli. It can merge 1000 ideas from cross-domains into three valid ones. Text-to-image tools can blend, combine, and generate unusual image options that would be otherwise unavailable.

Relying solely on AI may limit creativity. Other approaches, like biomimicry, can spark creativity. Nature is an exceptional designer and reminds us to “embrace with humility the idea that humans may not be the best designers in the world.” (Fast Company, 2023). The complexity of nature, plants, and animals is still largely unexplored, and here is where AI can be valuable, helping to analyze extensive data to deepen our understanding and solve complex problems.

We must view AI as a tool and collaborator, rather than a replacement for human skills and promote the value of originality and personal perspective in art. Education, transparency, and standards are essential in ensuring that AI is used ethically.

LLM Labs, which fuel the AI tool economy, have a big responsibility. The developers should enhance internal testing before releasing tools to the public so there is a chance for a democratic dialogue on how, what, and when to employ AI.

Creativity with AI could explode exponentially. AI tools are creative, easy to use, and affordable. Anyone can create with AI; the imagination and the artist’s aesthetic are the only limits.

“I find it so amazing when people tell me that electronic music has no soul. You can’t blame the computer. If there’s no soul in the music, it’s because nobody put it there.” (Björk, 1997).

YouTube explainer videos revolutionized learning for teenagers, and AI tools promise an even bigger leap: not just faster absorption of knowledge, but fundamentally transformed thinking through personalized learning, creative brainstorming, and democratized expertise.

Physical bodies make us human, and caring for all parts of our being—the body, mind, spirit, and soul—is essentially human and not something AI can replace. It is a special privilege to have a physical body with a range of experiences, pain, and love.

AI would not know how we physically feel or what we need. Humans are emotional beings with multi-faceted capabilities and multiple intelligences. The possibility of AI consciousness raises fundamental questions about what it means to be human and conscious. “What role do humans play in the big scheme?” The advancement of AI pushes us towards consciousness. This leads to multiple avenues for further research.

I will discuss ethical implications in a book called “AI Ethics for Designers”.

• How we got here
• Why AI suddenly became a hot topic
• Chapter Image
• The Best AI Tools
• The Top Ten AI Tools

• 1. ChatGPT
• 2. Gemini – formerly Google Bard
• 3. MidJourney (MJ)
• 4. Dall-E 3
• 5. Adobe Firefly (AF) & Photoshop
• 6. Adobe Express
• 7. Stable Diffusion XL (SDXL )
• 8. RunwayML
• 9. Kaiber
• 10. Topaz
• 11. More Tools Worth Checking-out
• 12. Video demo
• 13. Prompt examples
• 14. New workflows with AI tools
• 15. How to prototype with AI tools
• 16. Case Study – The Real Hero
• 17. Can you enhance your creativity with AI tools?
• 18. Recommendations
• 19. Final Words