Has generative AI become a tired trope already?

We’re two years into the gen AI wave and AI fatigue is spreading.

On my side what started out as initial fear for the future and impending doom that AI would bring, quickly turned into fascination of the new prospects AI could bring (for instance in the medical field).

I was inspired and thrilled about AI, but I’ve gradually started to feel… not so impressed.

Maybe it’s because the extent of AI usage I’ve seen on social media so far amounts to people firing a stream of consciousness into a chat bot to make inarticulate articles or blog posts accompanied by generic (and unprofessional looking) pictures.

New studies show that AI makes people underperform

We see it all the time with AI generated images. They are turning into generic (and sometimes worse than average) visual content – uninteresting and unengaging to viewers.

Now, new studies prove that point, much to the surprise of the researchers of the study. The use of gen AI can limit creative problem-solving in teams and make them underperform!

Sure, gen AI helped the workers in the study to avoid the most awful ideas in the problem-solving tests, helping them not flunk completely, but the use of AI like ChatGPT also led to more generic and mediocre ideas in the teams.

The problem: AI is struggling with emotions

Actually, it shouldn’t come as much of a surprise to us, since AI doesn’t perceive the world through a human body. We’ve known for years that micro-expressions are an essential part of visually decoding the human body language, and thus get a better comprehension of the feelings and thoughts of other human beings.

According to other new studies Gen AI is struggling to accurately convey other emotions than happiness. This indicated a gap in AI's emotional range, thus making AI generated images less valuable for professional use.

If you haven’t heard of micro-expressions the short explanation comes here: micro-expressions are facial expressions that occur within a fraction of a second. Those involuntary emotional leakages expose a person's true emotions and thoughts (like disgust, anger, fear, sadness, happiness, contempt, and surprise).

Read more about micro-expressions here>>

Has my love for AI faded?

As some of you might know (if you know me personally or you’ve been reading some of my previous articles both on my blog and here on Linkedin), I have been deep-diving into the world of AI for the past two years now to get a nuanced look on the prospects and possibilities that AI bring to animation and illustration, how I can use it to my advantage, and what I need to be aware of in future trends in my business and personal life.

I am even teaching AI classes from time to time (for instance for DMJX and Mediernes Efteruddannelse) and giving lectures and webinars on different topics of image generative AI. So, how can I even start an article like this?

I’m still reluctantly enthusiastic about the prospects of AI in the future (for instance in the medical and health industry). But we need to change our focus from perceiving AI as timesaving solution (that replaces humans) to a tool we can leverage to our advantage – on top of what we are already doing.

AI doesn’t make us work smarter or faster!

We are literally surrounded by AI and algorithms every day that make our life easier (for the most part). Think about all the Google searches you do a day, or how you use the spelling and grammar check in Word without even blinking. AI is with us everywhere (yes, some people even take their phone with them to the bathroom or sleep next to it).

But the new real-world experiments from ?Harvard Business Review show that even though ChatGPT’s ability to generate lots of ideas is boundless, the ideas themselves may not be so good. As the article points out:

“Common misconceptions about generative AI, problem-solving, and the creative process are causing workers and their managers to use the tools improperly, sometimes leaving them worse off than if they’d proceeded without AI input”.

Can AI save time?

Time is the ultimate currency! There is already the widespread misconception about image generative AI being a time saver for professional creatives. But often this is not the case.

According to many top managers, magazines and newspapers, we will all be able to generate usable and professional looking images and artworks in little to no time at all on platforms like Midjourney, Prome or Mage.

Many companies aim to solve GDPR restriction or copyrights disputes with AI generated content. Their hope is to cut down costs on photographers, animators, or graphic designers in the future.

However, from what I have learned so far in my work with image generative AI and on the courses, I’ve taught, AI is not a time saver at all. On the contrary. AI is hard to control and manage. It gives an output that varies much in quality (or lack thereof). And even though it is possible to edit images, video and text on AI platforms or in Photoshop, it’s a very time-consuming process – and the end results are far from perfect.

Sure, you may be able to create images or video that are 80% usable – and good enough for private use. But for working with clients? No.

AI makes us lazy

This statement is true for both people who work in communication and for the viewers and receivers of that communication. AI makes us lazy. We settle for less than good – because it’s easy. But we also start losing interest in the visualizations surrounding us if they look too perfect, soulless, or expressionless.

Just like industrial designer Ti Chang puts it in an interview:

“As humans, our eyes are drawn to the imperfection and the less flawless. So, when we see these images right now…I am turned off by them, to be perfectly honest. Because I don’t feel they were generated by a human; they were prompted by a human.”

For me personally, AI images have become this strange mix between paradoxically sharp and hazy at the same time. Like the filters you can use on Snapchat. Trained models of people look as smooth as Barbie dolls. Sure, it can still be efficient to use AI to generate sketches, style guides, preliminary ideas for animation characters. But it doesn’t save time.

What does the nearest future bring?

If you want to use AI professionally, you need to constantly learn about the new AI solutions out there. And that is quite a feat. For instance, Google has just launched its new Gemini ecosystem – with some minor bumps on the road (it turns our Gemini is too woke at the moment and is being readjusted).

Microsoft has fused ChatGPT into their new Co-Pilot software (making it possible for AI to search your laptop, researching, recapping, and making your work so much easier – like Apple always dreamed about with Siri).

You can learn much more about AI on Youtube. But beware, it’s a rabbit hole and you might spend hours and hours of learning stuff you don’t need and that gets you nowhere in the end.

How to use AI properly?

For now, the way I use AI is in the initial process, starting up my creative workflow. Just like I research the topic I’m illustrating or doing an animation about – by for instance Googling and reading up on information and studies, I also prompt visual output – to see what the different AI solutions comes up with. AI is an addition to my workflow. Not an oracle that can output finished results.

From there my company can create a visual style that fits my client or take some of the ideas that have arisen and fit them into the client’s business narrative or strategic visualization. Also, manuscripts can be checked for grammatical errors and better phrasing by AI. But I never fully rely on AI for creating anything other than drafts or preliminary sketches. Never.

How to get started?

So, perhaps managers should teach their teams to not rely fully on AI to do the hard work. In return managers should not expect their employees to solve their given tasks at a faster pace, just because they use AI in the process.

AI can be a tool in the creative process and creative problem-solving. But just like every other tool, you need to learn how to use it - with care. You can start by:

1. Creating highly specific problem statements to feed into AI as requirements.
2. Remember to include as much detail as possible (remember AI is not smart like a person).
3. Be specific about what you want AI to create for you. What should the outcome / product be?
4. Train your own bots and models. Generative AI systems lack the contextual understanding that professional people gain over years of working in their field and industry.