pull down to refresh
Someone will have to instruct these LLMs and verify what comes out. So I'd say there's more to learn, more to develop, more to jump into. And if you're still young, this will be much easier to do than when you're mid-career.
So I'd not fear for the youngsters, even if right now fomo is keeping purses a bit tight. I'm more worried that this kind of "resistance" will have the ultimate worse effect: first denial, then battle, then capitulation. Better to learn, even if you're skeptical... I'm skeptical.
I'm not too worried for mathmaticians yet. Copilot can't even consistently add numbers off a spreadsheet file yet.
CoPilot is the most retarded integration, it's clippy meets autocorrect. I am not sure if that is representative. And MS isn't generous with what they integrate in terms of models either.
Instead, GPT being trained to solve complex math problems has been a thing for nearly a year. I still don't get why they went that route but they did. I understand the fear. It's just not a rational fear imho.
I still don't get why they went that route but they did
Marketing? Makes for good stories every other month.
Coding was already taken by Anthropic, they had to find their own niche?
And it's probably easy to train them on this with the vast body of solved problems available in math olympiades and other archives?
Seems to me that that's as lazy as your average underskilled llm enjoyoor that simply has no idea what they are doing so they just yolo their time away. Its related to the AGI vision where it doesn't matter what it does because it will become sentient and succeed humanity, and we need to personally make trillions now so that we can maybe escape the fate of everyone else.
Except, no one believes that bs anymore. Which is probably why Anthropic beat OpenAI to now be the world's most overvalued startup.
Too much fearmongering. Not enough results. But I'm just a simple human being. I dislike liars. This means I will never be a billionaire, probably not even after hyperinflation.
Terence Tao, one of the bigshots in mathematics these days, is a big proponent of the use of AI in his research. Here a recent video/article on this: https://www.startuphub.ai/ai-news/ai-research/2026/terence-tao-ai-is-reducing-cognitive-friction-in-math
You can probably let your LLM summarize his thoughts on this scattered all over his blog. That guy is a machine in terms of scientific and scientific-adjacent output.
The issues they bring up about AI slop are real (e.g. how to deal with easy to generate but hard to verify submissions).
Have you seen an increase in AI slop in articles you had to review, in your field?
As a referee, it's not too bad for now in my field, the underlying science is still usually sound, even though I get annoyed at times, yet relieved that I don't have to read broken English anymore.
For editors, I heard, it's getting bad, and they can barely find referees anymore. But the system was already breaking before AI, AI is just accelerating it.
Honestly, all this hype just pumps up AI. Acting like this just proves AI is on the right track. It's straight-up free marketing for it!
It just sounds to me like they don't want anyone looking under their rug.
Mmm... all this discussion of values-alignment seems pretty squishy for mathematicians. A statement is either true or it's not, and a proof is either valid or it's not. Doesn't matter whether a machine did it or a human.
I wonder how commonplace this point of view is, or whether the majority of mathematicians are more accepting of AI.
The issues they bring up about AI slop are real (e.g. how to deal with easy to generate but hard to verify submissions). But the rest of the stuff like whether the tech companies' values are aligned, or what impact it might have on early career mathematicians, seems squishy.