@greyenlightenment's banner p

greyenlightenment

investments: META/FBL, TSLA, TQQQ, TECL, MSFT ...

3 followers   follows 0 users  
joined 2022 September 04 18:26:17 UTC

blog https://greyenlightenment.com/

Verified Email

				

User ID: 68

greyenlightenment

investments: META/FBL, TSLA, TQQQ, TECL, MSFT ...

3 followers   follows 0 users   joined 2022 September 04 18:26:17 UTC

					
				

				

				

				

				

					

User ID: 68

Verified Email

This does not quite fall under culture wars, but I thought it's still relevant given that it's a big story.

On October 6, OpenAI released a large batch of mathematical results, with more than 700 manuscripts covering hundreds of problems: Sharing AI progress in mathematics.

This has led to another round of controversy and discussion about AI making mathematicians obsolete or disrupting research programs that people have spent years working on. The Verge covers some of these reactions in “Mathematicians will need years to make sense of OpenAI’s latest drop”.

Many of the results are still being checked, but the release has called into question if mathematics can survive as a profession, or what purpose it serves if machines can outpace humans. Research programs have supposedly been gutted by these results. "Careers ruined," as the narrative goes. I think that is an exaggeration or overreaction. There are thousands of mathematicians and their interests are very wide ranging. The odds of someone's research program or problem being "solved" by AI are very small.

I don't see anything changing much. The status quo of journals and academia will remain intact even with rapid progress in AI, save for the occasional breakthrough in an unsolved problem. What is overlooked or ignored is the massive gap between cotemporary research-level math vs. LLM-produced math. A math PhD program, especially from a good school, teaches or guides future researchers towards productive areas of research that are of interest to the "math community," rather than results that are merely "personally interesting". AI cannot recreate such communities or networks.

AI is great for generating computation-laden results on-the-fly, but research-level math is more concerned with abstractions and relations between them, than crunching. I recall one example where ChatGPT’s Astra program tried to pass off a finite computational cutoff as a “theorem.” Um, you can't do that unless you prove that the bound is sufficient.

To test this, I produced some AI papers which I thought were great and novel, and the AI said so, and journals rejected them (although others are still pending at various stages of peer review). I had some knowledge in said area and could assess novelty, but still not good enough. I don't see laypeople doing better in this regard, although could be wrong. I ran my ideas through the Ai and got 12 papers out of it, which saved considerable time had I written them myself. It also found non-obvious connections and results. I am really satisfied with the results, but does this meet the bar for publication? Time will tell. I bet not.

Also, success is heavily dependent on the prompt. It's not as easy as blindly prompting something. I made a post abotu that https://greyenlightenment.com/2026/10/11/the-nuance-the-math-debate-is-missing/ I suspect many of these successful proofs were prompted by experts who knew what to look for, rather than just naively trying. What is touted by the media as an AI breakthrough, may involve surgically precise prompting in the background by an expert in said field.

Alluding to the "airplane meme," there is also survivorship bias: you don't see all the failed attempts with AI . Sometimes it works great, but other times hit or miss. Presumably , the "top unsolved problems" have been queried by frontier models and using large token budget, with few proofs to show for it (save for the Open AI "dump").

It would be cool to see AI truly shake -up math and breathe new life into ignored areas, but that is not going to happen, imho. But I do see AI leading to a 2-track system of decentralized preprints and GitHub repositories of AI-produced results, alongside established academia and journals. They will coexist, rather than the later being obsolete. Overall, mathematicians don't have to fear losing their jobs anytime soon (unless of course you happen to have the precise problem you are working on scooped by AI, and even then, your skills would presumably transfer to new problems.

yeah he's trolling. report him

think had trump not done the Liberation Day tariffs in 2025 and not attacked Iran. His approval ratings would probably be at 45% now

I mean interfacing with the real world . The containment breach during the Hugging Face incident was not that surprising and does not change my priors.

Sorta. A high GPA means you’re intellectually capable of absorbing large quantities of material. It’s more than just being able to function in a particular environment. If an employer sees that someone has a 4.0 in literature or philosophy, that signals they’re good at reading and making sense of large amounts of text. This is useful in a work setting when having to process reports, or make sense of disparate pieces of information. But the high GPA is also indicative of an ability to fucus and get work done, so there is an IQ and attentiveness component.

comedy is like the dream job because you get paid to tell jokes. there are no responsibilities. it scales well: you can go on the road and use the same material for years, with extra income from specials, albums and various residuals . The downside is misreading the audience. this is called bombing. Some might blame it on political correctness or an overly sensitive audience, but this is cope, imho. Your job is description is literally to make people laugh, not to rescue people from burning buildings or stuff like that . the onus is on the comedian to connect with the audience, and failing that, you no longer have a comedy career and someone who does make that connection will take your spot.

My p(doom) is still around 0%. The AI doomers have still not provided a convincing argument for how AI actually breaks containment. The most realistic disaster scenario would involve AI being put in control of infrastructure and then malfunctioning because it misread some signal.

It's amazing how trump is such a big public figure yet so little is known about him outside of that persona . What i have gathered is trump is lazy and delegates a lot, but this is not bad . He does the minimum needed , but this allows him to cover many areas, like being POTUS, media, and running his businesses. He does not care that much about the details of policy , compared to someone like Clinton. I think the only details trump truly cared about pertain to his real estate business.

the OP probably meant boots on the ground . the SCOTUS killed the tariffs anyway

many employers use the Wonderlic, which is very similar to an IQ test, in addition to checking for the usual credentials. But some may only use the Wonderlic. I agree IQ is way more efficient, but OTOH it does not test for diligence . getting a degree is a grind due to large coursework and having to show up , and many people do not finish for this reason .

I look it as, if I walked up to a woman on the street and called her that can I reasonably expect she would take offense? In this case yes.

the question is, would this has been discovered eventually anyway if not for AIAYN?

this was not news to to me. this is what I expected and i bought stocks in on the dip on April 2025 expecting him to do that

it was more like he would slap a high tariff and then back down , either giving some ultimatum or reducing the duty. The "taco trade" was buying the dip whenever he initially slapped a tariff in anticipation of him backtracking or extending a deadline, which he always did.

or even more derangedly, blames the USA.

would this be happening had Trump not attacked Iran in the first place ?

yup https://www.washingtonpost.com/national-security/2026/10/15/8-10-americans-oppose-live-streaming-execution-fort-hood-shooter-poll-shows/

this was not some change of heart or a legal obstacle: it simply polled poorly and would be unfavorable among independents. had it polled better i am sure trump would have signed off on it.

yeah but the sort of person who is diligent enough to get into an ivy would presumably like the high surface area of success of the consulting job

oh nevermind i misread that as a 1.33x bench-press

yeah as long as there is a gap, the universities serve a purpose. the college dropout rate still stands at 50% and many still take 5-6 years of a "4 year degree "suggesting it has not necessarily gotten easier. I think this "lowering standards at universities" is much more anecdotal. There are certain standards that degree-awarding colleges must follow.

I think they are both true. his book is right but at the same time employers derive value from degrees.

nice job on the bench press.

how is your blood pressure so low? mine was at 140 even at 180. congrats on weight loss . CICO is technically true in that sense that obesity reflects an energy imbalance, but also limited in the sense it's hard to predict how the "CO" variable will change.

this is a good point:

In 2025, a normal life makes you fat. It shouldn't take an extraordinary life to avoid being fat. And, for most of human history, it didn't. Sometime in the last few decades, something changed such that ordinary levels of exercise and satiety and willpower simply aren't enough anymore to avoid being fat. Since most people do not have the slack to redesign their entire lives around being thin, a realistic solution to the obesity epidemic needs to involve either identifying and removing the orange soda or inventing some kind of orange soda antidote. Telling people to just consume less calories than they spend is a useless distraction, like telling an adult who counts on his finger to just study harder for the AP Calc exam.

at least 75% of the US population is overweight through no effort. something is wrong. it's more than just "eating too much," as there is evidence people consumed similar or greater calorie counts hundred+ years ago without becoming as overweight. historically, when food was cut off during war or other famine , people rapidly starved to death , even on 1500 kcal/day like in Holland or Russia, yet we have entire Reddit subs where people cannot lose weight on 1200-1500 kcal/day (obviously, miscounting calories is an issue).

This sounds like a task for AI. I am sure someone has done a data analysis on this

Sor a second I thought this was Rutgers

Universities love to say they "teach you how to learn". In my mind, this is cope. There is no indication that they are making an effort to do this, nor that it is even possible.

Regardless of their purpose , whatever it is, employers sure want it.