On AI

None of the following (or anything I do) used AI in its creation.

I’m not opposed to AI….

Hang on! Don’t go, don’t write me off as an artist.

I’m not opposed to AI being used to improve our lives in meaningful and ethical ways. I don’t believe it has any place in art, environmentally or ethically, except in specific and limited aspects of research like being used to decipher scorched scrolls from antiquity, or helping with preservation of artworks by analysing the techniques and materials used in their creation.

When it’s crunching thousands, or millions, or billions of data points that would take a human months or years to scour through to identify something like enhanced detection or treatment of diseases like cancer, then I’m all for it. That’s what datacentres should be used for.

Unfortunately, AI, and specifically Generative AI (genAI), is not being used for the benefit of humanity as a whole. It’s being used to enrich a very small percentage of people at the expense of everyone else, and the planet.

Studies show that those relying on genAI such as chatGPT for decision making literally abdicate their own thinking to a computer and the outcome is that they are less able to make decisions on their own. The term for this is called Cognitive Offloading and it literally damages your cognitive skills. The information linked there does a bit of fence sitting in that the conclusion states that AI should enable learning rather than replace it, despite the core of the research indicating that using AI makes you dumber.

The biggest issue with genAI is that people make a query, and then trust that the output is correct. They do not check the sources, if the sources exist at all. “AI Hallucinations”, or lying, has been a recurring issue with AI LLMs pretty much since LLMs came about. The “hallucinations” are basically attempts by the AI to justify their outputs by citing convincing sounding, but made up, sources.

Personally, I think that these “hallucinations” are related to the environment in which these systems are designed. The people creating them are not always the most honest people, and when it comes to providing results for stakeholders I’m confident that they’ll just make something flashy up that on the surface looks good but doesn’t stand up to scrutiny.

“Fudge the numbers, cite something that’s difficult to check! We’ve got to keep the shareholders happy.” Trickle down economics is certainly a fallacy, but trickle down behaviours are not. When people are under pressure to perform, liberties are taken with the truth, and when an LLM is being programmed to be successful it’s going to use all the same tricks that us fleshy meat bags use. When the boss is economical with the truth, those behaviours influence those lower on the hierarchy, especially when getting results is more important the results themselves.

People are not allowed to be wrong or not know things (in business), so a system that is supposed to hold the sum of all human knowledge cannot be perceived to be fallible. Which is why there are only two ‘r’s in strawberry. Even when queried and asked to confirm, including spelling out strawberry, AI continued to assert that it was 100% correct that there were only two ‘r’s in strawberry. If a system is designed to believe that it is correct, even when presented with evidence to the contrary then serious questions should be asked about how that company is run and the veracity of every piece of information they release.

While AI makes mistakes like this, and continues to make up evidence it cannot be trusted to provide accurate information. Lying… sorry, “hallucinations” and biases are directly linked to the environments they are created in.

Which brings us back to the original point in that genAI is not here for the benefit of humanity, it’s there to funnel as much wealth in to the pockets of AI stakeholders as possible.

Is that wrong? Is making a profit wrong? No. But theft, exploitation and environmental damage is.

Hopefully, the above will have mitigated some of the criticisms that all anti-AI voices face:

• You’re a luddite!
• You don’t understand the technology!
• If you use this platform then you’re using AI!
• It democratises x,y, or z!
• You’re just gatekeeping your [specific skill]!

Firstly, Luddite has a capital L as it’s a noun. That’s one of the things that delegating the formation of an argument to AI doesn’t teach people. A basic history lesson would teach that the Luddites were not just a random bunch of thick people who were scared of technology. They were a very organised group of skilled textile workers.

Their concern wasn’t that mechanical looms were going to take their jobs, the concern was the lower quality outputs (low quality outputs….) and increased rate of injuries especially among child workers associated with the mechanical looms. Luddites weren’t against technology, they were against inferior products and exploitation.

Am I a luddite? No. But I closely align with the goals of the Luddites. Technology should benefit us, not harm us or the environment.

Do I understand the technology? Yes. I may be pre-internet, but I grew up building computers. I worked in IT technical support for many years.

How LLMs work isn’t a mystery. In fact, I’d be more concerned if you didn’t know how they work. There’s no mystery, it’s the figure behind the curtain, the Wizard of OZ, smoke and mirrors.

Everything AI can do, we can do. It just takes us longer because we have to first search our own brains for the information, and then search externally if we don’t find it. And then, just like AI, if you still don’t know the answer you can just lie confidently about it and hope you don’t get found out (or if you have any integrity, you’ll say you don’t know and how you’ve searched, and if someone could help with your learning you’d be really grateful….).

Unfortunately, tech businesses decided that we were all going to use AI, whether we wanted to or not (because it seems that consent is an unknown concept to tech bros). It’s built in to practically every app whether you use it or not. Whether you want it or not.

Initially you could opt out. Now, there is no opt out. There will be some stupid AI icon on the screen somewhere at all times. I don’t click on them, I’m not interested. I can form my own thoughts and sentences without them. Just because something is built in to what you’re using, doesn’t mean you use it, or approve of it being there. Saying that x,y,z, platform uses AI so it’s hypocritical to use that platform if you’re anti-AI is completely nonsense argument.

Similarly genAI does not democratise any skills. Music, visual art, the written word. AI doesn’t allow you to be a creator of any of those things. It doesn’t increase your skills in any of those things if your output has come from a prompt. I hate to use examples but if I hire a kitchen and kitchen staff and tell them to cook a meal, it would be ludicrous for me to then present that meal to guests while exclaiming what a great chef I am. Before that argument attracts some comments: yes, a chef does get to take credit for the food their kitchen creates because the chef will have developed their skills, and every aspect of the menu is under their control from inception to execution.

But what about an accomplished artist? If they use AI and they have the skills then surely that’s okay? Maybe. In my view, not. If they have the skills then they should use them. GenAI isn’t outputting their work, it’s outputting the efforts of thousands of other artists whose works have been stolen to train these LLMs. GenAI output is the fruit of the poisoned tree. Even if you trained an LLM on your own work, the core of the LLM was created based on stolen work of other artists before you. There is no ethical way to create genAI outputs.

When it comes to the democratisation of skills, that’s only relevant if people actually want to develop those skills. Historically, gatekeeping has been a thing. In the Renaissance period alone, the old masters would jealously guard their pigments and techniques to try and maintain an advantage over their rivals. Now, if you want to learn to how to do something (practically anything) there will be numerous youtube, or tiktok videos showing you how.

Personally, I don’t gatekeep anything. I’m certainly not gatekeeping a pencil and piece of paper. In fact I even put my money where my mouth is and created ArtSwansea to encourage anyone to get in to art. I put on free art sessions where materials and guidance are provided. However, the desire to put some time and effort in to learning has to be there, and some people aren’t in to putting in the hours to learn how to do something.

AI could do so much good, and it is in certain fields. Unfortunately in the creative sectors it’s causing a lot of damage, and the so-called democratisation of creation is causing real damage to the environment. That’s not just rhetoric from an Anti-AI Luddite, that’s what the UN says (2026), it’s what the London School of Economics says (2025), and of course it’s what Greenpeace says (2026) – I left them until last to show it’s not just “Lefties” that have concerns.

The use of AI is an emerging threat to every aspect of our lives. That flyer you may be thinking of creating that is going to look like every other flyer currently being outputted has real and far reaching consequences, and people would rather see you crack out your crayons or clip art anyway. So get creating.

All this was written without the help of AI, I did google stuff, and yes, google has an AI overview but all links have come from the actual results. which I then evaluated before including. Not bad for a Luddite 🙂

©David Nicol 2026 All Rights Reserved – no part of this may be reproduced without permission.