Über 150 Mathematiker warnen Regierungen davor, dem Hype um KI zu „glauben“.
https://www.yahoo.com/news/science/articles/over-150-mathematicians-warn-governments-100000243.html?.tsrc=daily_mail&segment_id=DY_VTO_50_Supernova&ncid=crm_19908-1475736-20260607-0–A&bt_ee=MEbzd%2FT3CK9hBFZUv6x%2BXxtzL%2B1%2B%2BKmVwclWdPE4ceWgse1VAnaUOsvcOk%2BPZovJ&bt_ts=1780835533932
28 Kommentare
Breaking: Anthropic announces AI will replace mathematicians.
Trump asks: „Never heard of them, what are mathematicians?“ /s
All great… and then:
> „Mathematicians who never intended to contribute to AI development are having their work used for this purpose without their consent,“ Leiden University anthropologist of AI
Duh, math is math, you can’t gatekeep it.
I’m not sure people understand AI for high level skills. Hear me out please before you just downvote. If you need AI to solve „what is the diameter of a pizza“ or something, sure. When you’re dealing with complex theories you need to actually understand the underpinnings of those theories to tell it its stupid.
AI to me just accelerates the dunning Kruger effect. People who don’t know how to figure out the diameter of an 18″ pizza are not going to be the ones solving complex mathematical problems, or a better airplane wing design. You can’t just prompt „solve this unsolvable problem“ and expect it to work.
It’s like a new employee, it’s going to be like „uhhhhh this?“ but you have to be smart enough to say „you did it wrong, it needs to be like this“ until it figures something out.
Like any tool, buying a CNC machine doesn’t make you a machinist. You can do a bunch of stuff that’s cool, but you still need to understand the limits of the machines for it to be useful.
It doesn’t replace education, it doesn’t replace mathematicians.
I’m not advocating for AI, I’m just trying to explain how it works and that beyond basic questions (like is an 18″ pizza more pizza than 2 12″ pizzas) you still need to know what you’re talking about about.
My company forces me to use it for development and I spend 1/2 my time arguing with it more than it’s able to build me something useful, but it can help me on that last bit.
EDIT: lol an immediate downvote as soon as I posted this lol.
EDIT2: I’m not really going to defend my position here. I’m not speculating about what AI „CAN“ do in the future. I“m replying to an article. For all I know AI is going to destroy the world. Dunno. Not what I’m speculating on here. I’m talking about AI today, and what AI can do, TODAY and the foreseeable future. This reminds me more of the hype of Bitcoin which is also worrying that people don’t understand its current fences.
I don’t get it. Or I’m missing something. What products have been improved with AI?
I see tons of cuts to employees raising stock valuations, but I have yet to see a single product improve. Amazon, the marketplace has gotten worse to navigate, deliveries now come in a way less organized fashion (multiple deliveries in a day instead of bundling them into one box etc)
Google search is 1000x worse than it was in just 2018.
Like what is AI improving besides stock prices?
Feels like the uranium fever from the 20s where they just shoved radioactive isotopes into everything because it was trendy
>It’s also a pertinent reminder that AI models are being trained on cutting-edge research, often without sign-off from the original authors.
>“Mathematicians who never intended to contribute to AI development are having their work used for this purpose without their consent,“ Leiden University anthropologist of AI Rodrigo Ochigame, who helped draft the declaration, told *Scientific American*. „I think that’s a deeply concerning situation.“
Is everyone’s IP just fair game for AI training now?
lmao what? AI has already helped with many new discoveries in mathematics. https://mathoverflow.net/questions/502120/examples-for-the-use-of-ai-and-especially-llms-in-notable-mathematical-developme
I’d say it’s 50% too much hype and 50% not being prepared enough to how it’ll impact society.
I think the article is nice. It says AI managed to solve 2 problems. I’m sure it can manage to solve much more in the future. Maybe in the next years, 100 similar problems will be solved with AI. And I’m also sure they already tried current models on thousands of unsolved problems and the AI completely failed to solve them and will continue to fail.
So yeah if you say it can’t solve anything, it’s wrong, and if you think it’ll solve everything, well it’s also wrong. And obviously because AI is a very good bullshiter, you’ll always need humans to triple check what it says.
Climate researchers are like *“First time?“*
Trump: I’ve never heard of these magicians
AI solving complex math problems doesn’t necessarily turn into human discoveries. Or discoveries that benefit humans. Certainly no one’s furing the math department and replacing them with AI anytime soon.
Turning over the military or worse the surveillance of the general public to AI, isn’t genius, it’s merely the act of humans deciding to avoid being implicated if things go wrong.
Finally, if any actual business claims it has replaced staff with AI „Agents“ but customers should still pay the same price for its product or service, 3 minutes after that, a competitor with better AI Agents will undercut that price. Until the final winner will be a teenager in Mumbai who has the best AI Agents, offering the better product or service for „free with ads“.
AI is a TOOL that can substantially increase the productivity of humans. It should be looked at as a really good Word Processor. Nothing more.
I think most are beginning to realize that the AI pumping has been done by those with perverted incentives and that reason is preferable to bullshit and hype.
Somewhere, Flavor Flav is smiling.
IMO a lot of these AI CEOs come off as scammy. Like you can tell they don’t actually believe what they are saying.
Breaking: Anthropic says human race will be discontinued, more at 11 – 👨🏽🦳
Lol
The hype is real with AI as they all try to make any $$$ off of it to control market share. But the reality is that we are still years if not decades away from what they promise as far as capabilities.
I dont doubt the generational shift that AI will bring but I think its going to out more like the way the internet did.
Decades of obscurity and gatekeeping, then the adoption of crude models with a bubble when it becomes unsustainable. Then the practical usage that ruins society as we know it.
AI is logic bound so its logical is sound. The problem is the data it’s trained on is assumed to be correct. If AI were developed in the 1940’s it would support segregation and lobotomies as consensus positions. What do we assume to be correct today that is actually erroneous? These errors will compound and propagate as they are masked by fluent output.
AI is proving to be Dunning-Kruger wrapped in an IPO.
if i’ve learned anything from politics in the past decade, it’s that scientists or experts advising a politician about something is the surest way to make them fully commit to whatever goes against that advice, because then they can tell themselves that they are smarter than the experts.
Because AI is a glorified web scraper. That’s all it really is. It takes extant ideas and mishmashes them into “original” work
It’s not the hype that politicians believe, it’s the 💵
PSA: Every major AI post gets astroturfed by pro-ai bots deployed by Reddit and their investors, as well as interested independent third parties.
Tech bros bought trump, trump invest ls U.S. govt fund into ai. We’re stuck financing this crap.
Its genuinely just dangerous and potentially species ending if you look at projected further implementation of it
See the same kind of overhype with medicine- loud verbal people report success with ai being “the only one “ to successfully diagnose my condition, but the problems and inaccuracies remain troubling when applied consistently over clinical contexts. People shouldn’t consume and internalize so readily the hype these companies are putting out .
“There is currently a strong commercial incentive on the part of the technology industry to overstate the capabilities of their products.” Maybe in their marketing material, but there are disclaimers in small print on ChatGPT, Gemini, etc., and I’ve seen them make mistakes first hand.
Maybe it’s part of the same mentality as survivorship bias, but my job should be right in the crosshairs of AI. It’s supposedly very easy to automate. But it just never happened. I schedule things for government employees. I’m certainly not looking forward to it, but I’m starting to wonder if this technology exists or not so I can prepare accordingly.
They should also warn governments that 911 ain’t comin in yo town.