Did I just write almost 5000 words about why generative "AI" should be resisted, going through every single one of my reasons?
Yes. Because I need this article to exist so I can point people to it.
Enjoy and share widely.
https://estherweidauer.com/articles/why-resist-generative-ai.html
This article has been living in my head for months now, often preventing me from focusing on anything else.
I'm so glad I got this out finally.
@aesthr under politics, it might be worth calling out how the training data explicitly reinforces biases already present leading to additional sexism and racism for example. I think you alluded to it, but if I wasn't thinking about it I probably would've filled in the blank with something else.
Otherwise, it's super helpful. I'll definitely be sharing it with people I know. I also appreciate some of the suggested readings. I'll have to check those out.
I agree. LLMs are terrible at predicting boring, reasonable behavior or outcomes. If asked how humans would respond to a friendly visitor from another planet (like Rocky from Project Hail Mary), an LLM recently responded in great detail how humans don't respond well to unfamiliar outsiders and would poke, prod and ultimately kill the alien based on ample evidence from human history.
In a way, you can't blame the LLM for projecting that from our sordid history but, on the other hand, it doesn't take into account that most history is written about the dramatic, violent highlights of the past and not the common acts of kindness and acceptance that rarely get proportional representation.
So, ask an LLM about human behavior and it will confidently tell you how people rape, pillage and exploit everything they can get their hands on which is both accurate while also not showing the whole picture.
Oct 20, 2026, 2:15 PM@aesthr Good job! Thank you.
@aesthr Reading this now. Thank you for writing and sharing this.
@aesthr that’s half my reason for writing, so I don’t have to repeat myself, lol.
@aesthr Excellent stuff, thanks for writing it!
Oct 20, 2026, 7:05 AM@aesthr
Great piece!
Thank you for writing and sharing!
@aesthr Really good and well-sourced piece, thank you for writing and sharing it!!
I spotted two typos, in case it's useful:
"generative pe-trained tranformer" -> pre-trained?
"Longermism" -> longtermism?
Now the question is how do we go through people who have already caught the genAI virus? I've been sharing these arguments with some that I know and it's just not going through...
@elduvelle in my experience you need to find a thing they care about and then tell them about how "AI" damages that thing.
For example I've had success with a data-protection angle recently for people in the medical sector who do really care about the confidentiality of their clients
@aesthr
Thanks for taking the time to write this!
@elduvelle
I also take the approach of focusing on a thing they care about. The feeling "it's just not going through" is shared! Some reply "I know you are right, but I'll continue to use it nevertheless". But it is necessary and useful to explain once and again; like planting a seed, the effect is not immediately visible, and often it simply won't germinate. But sometimes it works.
A minor complaint on "LLMs are simply not capable of producing new knowledge". In mathematics, new knowledge is certainly available through the use of LLMs, and technicalities or stating that it was implicit in the sources won't convince my colleagues. What I do is try to show that the actual cost is overwhelmingly bigger than the benefit.
Could explain?
... In mathematics, new knowledge is certainly available through the use of LLMs...
@Davide_Sandini OpenAI (and also Anthropic, to a lesser extent) have released "drafts" of proofs or counterexamples of hundreds of previously open problems, often with a formal verification. It is obvious when reading the drafts that there was no mathematician expert writing them (they show no understanding of the issues and the importance of each part) but with effort they can be understood and most of them are correct. This means that we now know the truth status of certain questions that have been studied for decades without success, and this through the use of LLMs.
I still refuse to write that "the LLM proved it". There has to be a human prompting it cleverly. But the LLM was essential.
@aesthr thank you for writing this!!!
I’m not quite through reading yet, so maybe you’ll pick the topic up again later, and filling it in:
I was surprised to not see a sentence to the effect of “And we can (and do) change them.” at the end of:
> What we do know is that the human mind is not a set of computer instructions. Neither is it a disembodied entity that’s just piloting a body. We are embodied beings and our bodies are an integral part of who we are.
@danyow that's not in there because it's not really relevant here. The point is that we ARE our bodies and not just HAVE one.
@aesthr true.
@aesthr good post covering a lot of ground, thanks. I liked the calm tone and the references.
Maybe I missed it on a first reading but I didn't see any discussion of the copyright theft issues and the effects on writers / artists / musicians.
There are some parts where I would have been more strident if I was writing it, but that probably just shows my biases - as I mentioned earlier, a good calm tone.
Thanks
@imcdowall yeah I might add a section about the stolen training data but I have to think about it some more in order to not just accidentally write something pro-copyright
@aesthr @imcdowall Moral rights might be the angle you're looking for: https://en.wikipedia.org/wiki/Moral_rights
Feels like it should be possible to argue for the basic theory of a creator's moral rights over their work, without propping up any particular implementation of copyright law.
(As an author in the U.S., I rely upon U.S. copyright law for my livelihood, but at the same time decry the deep flaws in its implementation and the inequity in its selective enforcement.)
@aesthr thank you so much for putting in writing what I think about daily and in such a great way. I also will point people to this article, because it is a beautiful summary of everything that's wrong with this hype.
Thank you, very much appreciated.
Oct 20, 2026, 8:24 AM@aesthr Thank you for writing this and sharing it. It´s bookmarked and will be put to good use.
I just wish I could have it printed on a t-shirt I could point to, when some guy (it´s always a guy somehow) starts rambling about AI.
Danke!
@aesthr this is an excellent summary of the issues and much more eloquently stated than I could put it. I'll share with colleagues
@aesthr This is very, very good. Thank you.
Oct 20, 2026, 9:56 AM@aesthr I've been having the intention of writing this article for a long time, actually have a preliminary version.
Yours is so much better written, more comprehensive and better sourced than mine would. Thank you so much.
Would you be OK if I translated it to Spanish and posted in on my blog at some point? Referencing you of course.
@delcano That would be awesome, yes! :)
Oct 20, 2026, 3:38 PM@aesthr This is my first iteration, still subject to review by other people: https://blep.org/por-que-resisto-a-la-%C2%ABia%C2%BB-generativa/
@delcano btw, if you want to use the same cover image, feel free to copy it over there too
@aesthr thank you so much for this! An article like this was on my to-do list for way to long and now I can happily point people to yours 🥳
@leah nice! I'm glad it's useful to people
Oct 20, 2026, 10:27 AM@aesthr A good read, thank you! I will definitely mention the Pohl bowdlerized summarization experience to library users - very tangible illustration
Oct 20, 2026, 10:35 AM@aesthr One thing though - as I understand it, the Character .ai bankruptcy isn't a "real" bankruptcy caused by a lack of customers/investors, but rather a strategic move to transfer much of the company's IP and staff to Disney while leaving the lawsuits and liability with the empty husk of the original company.
https://www.wired.com/story/disney-asked-characterai-to-go-bankrupt-for-a-get-out-of-lawsuits-card/
Well worth a boost, thank you so much for writing it 🙏
@aesthr Thank you for this! Very helpful :)
Maybe i misunderstand something here:
"A more implicit example could be the steam engine. It’s not just an invention that happened in a vacuum. Without the political and economical framework of the fossil fuel industry it can’t operate and would lose its entire reason for existing."
'steam engine' needs fossil fuel industry?
Should it say ´combustion engine' or smthg?
Oct 20, 2026, 1:50 PM@cagraphics Steam engines require coal (a fossil fuel)
Although you can load the firebox with wood, it doesn't burn hot and long enough, meaning you can't carry enough fuel on the train for your journey without sacrificing a prohibitive amount of passenger/cargo capacity
@ben @cagraphics exactly.
Technically you can run a steam engine on wood (which still means it'll require a foresting industry) but to make it efficient in an industrial application, fossil fuels are required.
And modern natural gas power plants are also steam generators. Running those on wood would be extremely difficult.
Oct 20, 2026, 4:10 PMThe idea of wood-fired power plants reminds me of this: there was talk of using chicken droppings to help fuel a power plant, though we've recently decided not to - https://www.bbc.co.uk/news/articles/ce85dq9qvvlo
Oct 20, 2026, 10:36 AM@aesthr Esther, this is such a good article.
Thank you for this.
@onepict thanks :)
@aesthr Good for you.
Me too. Every day I write something. Not to publish, but to order my thoughts away from chaos. I’m at currently at, if you support any AI you’re a climate change denialist.
Sorry if this seems like a thread hijacking.
@aesthr Thank you for writing this. Most of the same points have been kicking around in my head around this, and it was cathartic to read.
@aesthr Great article! Thank you!
Oct 20, 2026, 11:06 AM@aesthr Thanks for writing this, because now I can point people at it instead of my ever growing messy link collection. Given the breadth of the subject, it's actually very succinct. :moomin_sparkles:
@aesthr After greatly enjoying this (I am completely aligned with your stance on GenAI) and boosting it, I’ll put my editor cap on and list a couple of typos I spotted:
Find: pre-trained tranformer
Replace: pre-trained transformer
Find: this has it’s limits
Replace: This has its limits
Find: one it is made
Replace: once it is made
Cheers! And as always, hashtag FuckAI.
@morrick Thank you, all fixed :)
@aesthr I don't agree with all of the article, but it's a really good one!
"[...] And once you’re using it for something that you couldn’t do well yourself, you have no way of validating its output, rendering it very much not useful while still appearing as if it is."
So much this! There was a quote I've heard early:
"It's interesting that LLMs make so many mistakes in an area where I'm a specialist, but are perfect in everything else!"
@aesthr Thanks for writing this down! I enjoyed some analogies, like the angle grinder vs fire department!
Small typo: The "b" in "but since it’s electricity ..." should be capitalized.
@blinry Thanks, fixed it
@aesthr great article, thanks for expressing this so well into words! 👏 I would love to read your thoughts on the recent Torvalds statements about AI: https://lore.kernel.org/all/CAHk-%3Dwi4zC%2BZe8e%2Bp3tMv8TtG_80KzsZ1syL9anBtmEh5Z40vg@mail.gmail.com/
@doublebyte I'm not familiar enough with Linux kernel development to talk about that more than a few surface-level opinions maybe.
@aesthr thank you so much for this
Strength and solidarity to you 💜
@aesthr I'm halfway through this and it's an excellent read. The middle part makes good, valid arguments against what AI proponents are saying. I am particularly offended by the claim that folks who do not dive into AI are Luddites. I made a living as a tech writer for more than 20 years and am in favor of tech that can actually improve our lives. I don't see that with what people are calling AI.
THIS!
Thank you.
I'll have to memorize some of these points for future discussions.
(Minor nitpick: double "now" in "see large companies now restrict their employees’ use now")
@Yendolosch fixed it :)
@aesthr Thank you so much, it is really good to have this all in one place/article.
A small typo: last par in Harms - Environmental, "getting worse than in already is": in -> it
As a good resource for the Fascism part I can recommend also the writings of @RainerMuehlhoff, specifically his book "Artificial Intelligence and the New Fascism": https://rainermuehlhoff.de/en/KI-und-der-neue-Faschismus-Reclam-Buch/
@aesthr I cant quite decide what i should think about the "here to stay" point. Agree, take it literally, or reject it? When (not if) the bubble bursts and VC runs dry, there will be no funding to train new models, so we're stuck with the models that exist up to that point; that's the "literal" reading. Could also be that people eventually get sick of those frozen models and they vanish over time, that'd be the "reject" reading. My biggest fear, however, is the "agree" reading:
@aesthr Chef's kiss!
@aesthr next month your article will be included in their new models....