The anti-AI thread

[Kaido]

And remember Windfarms are bad:

https://edition.cnn.com/2025/12/22/climate/trump-offshore-wind-suspension-virginia

Also a Danish company partly owned by the Danish state is heavily involved here, so Greenland preasure is also suspected.

Denmark has some of the cheapest industrial electricity in Europe, >80% renewable.

bruh 😭

The Virginia offshore wind farm is 60% complete, according to a recent state report, and would deliver more than two gigawatts of energy to the grid — enough to power around 660,000 homes.

 

[KMFJD]

do techbros ever think these things through?

AI data centers may soon be powered by retired Navy nuclear reactors from aircraft carriers and submarines — firm asks U.S. DOE for a loan guarantee to start the project

This is a much cheaper and faster way to get nuclear power.

www.tomshardware.com

 

[Kaido]

do techbros ever think these things through?

AI data centers may soon be powered by retired Navy nuclear reactors from aircraft carriers and submarines — firm asks U.S. DOE for a loan guarantee to start the project

This is a much cheaper and faster way to get nuclear power.

www.tomshardware.com

Hey so

FDA issues recall for frozen shrimp that may be contaminated with cesium-137

The FDA announced a recall of frozen raw shrimp because of potential exposure to a radioactive isotope.

abcnews.go.com

 

[Kaido]

Mark my words, AI slop will be the Plague of 2026 🤮

https://www.reddit.com/r/CookbookLovers/comments/1pwepxl/ai_generated_cookbooks_from_amazon

 

[Kaido]

This is worth the full ~20 minute watch:

 

[SteveGrabowski]

Wanted to see if ChatGPT could be useful for something mathematical so asked it to prove every Lorentz transformation could be expressed as the composition of a boost and a rotation since I don't have a proof of the result in any of my physics books. It spit out an answer telling me what baby's first Lorentz transformation was, asserted the result, made a handwavy argument after asserting, and then asserted the group of Lorentz transformations was the product (guessing it meant a semidirect product) of the orthogonal group and the group of boosts, which is bullshit because the boosts don't form a group. Then it asserted it proved the result. A lot of talking loud and saying nothing in ChatGPT's answer.

 

[SteveGrabowski]

Tried a much easier math question: prove the covariant derivative transforms as a tensor. It actually spat out a mostly correct argument, but left out a crucial part that changes the sign in one of the terms in the transformation of Christoffel symbols to show it cancels out with another term related to how partial derivatives of four-vectors change under coordinate transformations. So O for 2, first question a difficult one, second an easy but tedious one you'd think an AI would be able to easily do.

 

[lxskllr]

One of the big problems with AI is the answers can look pretty good with a cursory look. Really analyzing what it's given you reveals it's bullshit. Makes it dangerous if you're doing anything important. Bad data can get by you, and if it doesn't create a disaster, it'll at least make you look like a dumbfuck. In the time it takes to check it properly, you could've just done it yourself.

edit:
It's also random chance. When you have a dumb intern, you can get a feel for their strengths and weaknesses, and work around them. AI is a random results generator. Might be perfect, might be bullshit. Flip the coin and see how lucky you are.

 

[[DHT]Osiris]

Mark my words, AI slop will be the Plague of 2026 🤮

https://www.reddit.com/r/CookbookLovers/comments/1pwepxl/ai_generated_cookbooks_from_amazon

View attachment 135680

See this is something I find hilarious, bc I use AI to create/revise recipes all the time, including baking. If they'd actually verify the recipes in the book they'd be fine.

 

[SteveGrabowski]

One of the big problems with AI is the answers can look pretty good with a cursory look. Really analyzing what it's given you reveals it's bullshit. Makes it dangerous if you're doing anything important. Bad data can get by you, and if it doesn't create a disaster, it'll at least make you look like a dumbfuck. In the time it takes to check it properly, you could've just done it yourself.

edit:
It's also random chance. When you have a dumb intern, you can get a feel for their strengths and weaknesses, and work around them. AI is a random results generator. Might be perfect, might be bullshit. Flip the coin and see how lucky you are.

I remember seeing a page that had solutions for a lot of interesting math and physics books and among them I found Lang's Algebra, which is a really fast paced and difficult first year graduate text in algebra I spent years writing solutions up for fun in spare time at work or I'd read a problem and think about before bed, in the shower, on the commute, etc and then try my ideas out later when back at home. So I started comparing because I like to see how others solved problems in this very difficult book and it was a bunch of AI generated bullshit that just restated problems, gave simplistic examples instead of the asked for proofs, and asserted very nontrivial results or things that were flat out wrong. Yeah if you didn't know the material you might think oh this AI knows its shit but it was always just quantity with absolute zero quality in those solutions. I can't believe this is the tech that's supposed to replace human ingenuity.

 

[Kaido]

See this is something I find hilarious, bc I use AI to create/revise recipes all the time, including baking. If they'd actually verify the recipes in the book they'd be fine.

Oh sure, I use AI in a lot of food-related activities successfully:

1. Calculating macros for recipes
2. Guesstimating macros from a photo
3. Generating shopping lists from recipes I pick
4. Creating graphical enhancements for my smarthome system's meal-planning tool (see below)
5. Adjusting recipe yield (i.e. half or double)
6. Doing baker's math (ex. sourdough hydration percentages...I have math dyslexia lol)
7. Calculating schedules via timers & alarms (no-knead, sourdough, dehydration, sous-vide, long smoke jobs, etc.)

And so on...but creating a fake picture to represent a fake, untested recipe in an unvetted book is just straight-up lying lol. I mean, I do a LOT of AI-enhanced photography, so there IS a place for it! For example, I have a custom smarthome system that has a meal-prep system on a 50" TV at my dinner table (4K is $178 at Walmart, haha!) & I use Nano Banana Pro to turn cell phone pics into restaurant-quality images for me to scroll through when I'm doing my meal-planning, My amazing iPhone pic:



Prompt:

Convert this image into dark-mode food photography. Moody lighting, soft directional shadows, high contrast, rich colors, restaurant plating, 50mm shallow depth, 45-degree view, matte texture, minimal highlights, similar style to high-end recipe galleries.

Et voila:

 

[IronWing]

Google Scholar is not only making up citations to non-existent papers, it is now fabricating entire scientific journals. As fake citations spread among student papers, Google Scholar is eating its own bullshit.

 

[Kaido]

Google Scholar is not only making up citations to non-existent papers, it is now fabricating entire scientific journals.

Will only get more prevalent:

Journalist Caught Publishing Fake Articles Generated by AI

Individuals keep claiming to be freelance writers and pitching dubious, AI-generated pieces to unsuspecting editors.

futurism.com

https://www.reuters.com/technology/artificial-intelligence/ai-hallucinations-court-papers-spell-trouble-lawyers-2025-02-18/

AI hallucinations: ChatGPT created a fake child murderer

ChatGPT created a fake story about a Norwegian man, claiming that he killed his two children and went to jail. This never happened

noyb.eu

 

[Kaido]

Hack Reveals the a16z-Backed Phone Farm Flooding TikTok With AI Influencers

A hacker gained control of a 1,100 mobile phone farm powering covert, AI-generated ads on TikTok.

www.404media.co

https://twitter.com/x/status/2002029126568292741

Slop avalanche incoming!

https://twitter.com/x/status/1894568751296618976

 

[Kaido]

Dang:

Instagram chief: AI is so ubiquitous 'it will be more practical to fingerprint real media than fake media'

Adam Mosseri offered a notably candid assessment on how AI is upending Instagram.

www.engadget.com

 

[IronWing]

Defense lawyers are in for a future of fun and profit. AI generated police report of a bodycam recording picked up a TV show in the background and faithfully included that someone had turned into a frog during the encounter. While the officer caught and corrected this one, what are officers missing? When the report is submitted as evidence doesn't align with the bodycam recording, I suspect the judge will not be amused.

Heber City Police Department test-pilots AI software - Park Record

On Oct. 21, Heber City Police Chief Parker Sever appeared before the City Council and informed them that a police officer had shape-shifted into a frog.

www.parkrecord.com

 

[Kaido]

Defense lawyers are in for a future of fun and profit. AI generated police report of a bodycam recording picked up a TV show in the background and faithfully included that someone had turned into a frog during the encounter. While the officer caught and corrected this one, what are officers missing? When the report is submitted as evidence doesn't align with the bodycam recording, I suspect the judge will not be amused.

Heber City Police Department test-pilots AI software - Park Record

On Oct. 21, Heber City Police Chief Parker Sever appeared before the City Council and informed them that a police officer had shape-shifted into a frog.

www.parkrecord.com

Yeah we're hosed

 

[Kaido]

More than 20% of videos shown to new YouTube users are ‘AI slop’, study finds:

More than 20% of videos shown to new YouTube users are ‘AI slop’, study finds

Low-quality AI-generated content is now saturating social media – and generating about $117m a year, data shows

www.theguardian.com

Low-quality AI-generated content is now saturating social media – and generating about $117m a year, data shows

 

[Kaido]

 

[mikeymikec]

Filing under "you can't make this shit up":

No, Grok can’t really “apologize” for posting non-consensual sexual images

Letting the unreliable Grok be its own "spokesperson" lets xAI off the hook.

arstechnica.com

 

[HeXen]

The only thing I like ai for, is for creating alternate universe scenerio's with historical events and I find the results insightful and interesting. Usually it's old consoles like "What if Atari released the 7800 with *** CPU and a 4 button controller shaped like PS and just weird things like that. Or what if TG 16 wasn't released in America and NEC released the SuperGrafx instead and I'll change out hardware and even pack in games just to read about how it changes the pricing etc. I always wondered how much wholesale prices were on chips back then.

 

[Kaido]

Welcome to 2026

https://www.reddit.com/r/aivideos/comments/1q3nbty/somebody_that_i_used_to_know

 

[Kaido]

The only thing I like ai for, is for creating alternate universe scenerio's with historical events and I find the results insightful and interesting. Usually it's old consoles like "What if Atari released the 7800 with *** CPU and a 4 button controller shaped like PS and just weird things like that. Or what if TG 16 wasn't released in America and NEC released the SuperGrafx instead and I'll change out hardware and even pack in games just to read about how it changes the pricing etc. I always wondered how much wholesale prices were on chips back then.

You know...you should have AI write out those stories & post them in audio format on video social media as reddit-style stories while an obby is played LOL. These are the new "read the back of the cereal box while eating breakfast" & are getting zillions of views haha:

 

[biostud]

do techbros ever think these things through?

AI data centers may soon be powered by retired Navy nuclear reactors from aircraft carriers and submarines — firm asks U.S. DOE for a loan guarantee to start the project

This is a much cheaper and faster way to get nuclear power.

www.tomshardware.com

They don't need to, they just ask AI

 

[nakedfrog]

You know...you should have AI write out those stories & post them in audio format on video social media as reddit-style stories while an obby is played LOL. These are the new "read the back of the cereal box while eating breakfast" & are getting zillions of views haha:

Actually, no, the thing for decent people to do is not engage with it for pointless frippery.