Ves

I think people still try to send a message with songs (at least sometimes), but there are so many genres and people have really fractured into listening to things from the genres they like best. I saw Weird Al say this; he said he makes parodies so rarely now because everyone has kind of fractured into their own musical bubble. Apparently (I wasn't around for this), in the 80s an 90s things were very different: tons of people watched MTV and everyone was aware of the top songs on the Billboard Hot 100. It seems like you could push an idea to a much larger group of people if this was the case.

I think this isn't a very clear-cut situation with the information that is publicly available, it seems mostly like a breakdown of communication to me. You can read Tristan's statement here, and OpenAI's response here. In my opinion, there was some skullduggery by OpenAI in that they heard a rumor that these two collaborating mathematicians (one of which works at Anthropic) had made a breakthrough in this Euler problem, and then they decided to drop a ridiculous amount of money on Navier-Stokes as a result. It seems that the resolution of Navier-Stokes use different that the method that Tristan and Levent used to construct finite time blowups of Euler equations (something about "forced" vs. "unforced"; I don't know much about the field of PDEs in general), but I sort of wonder if OpenAI just got lucky that their model took a different route than Tristan and Levent. Tristan sort of jumped to conclusions that OpenAI had accessed his conversations with Codex, for reasons that seem reasonable in his account of what happened. There was also some pettiness with OpenAI not wanting Levent to be an author on the paper because he worked at Anthropic. I have a pretty low opinion of Sebastien and his colleagues after seeing some of them mock a heartfelt response to Levent on X, but whatever. The effect of this drama that I see is that mathematicians are concerned that OpenAI is training models off of their conversations (it is, unless you "opt out", who knows if that actually does anything), which would result in their arguments being plagiarized if AI regurgitates them in future solutions. This is a general issue with AI in math: it doesn't know who was responsible for the data it's trained on, so it can't cite any sources properly.

Yes. There is so much published mathematics that each person only knows a tiny fraction of what exists, even in their own field. I mean, I study algebraic geometry, which was revolutionized by Grothendieck in the 60s. He published a ton of papers laying out the new foundations of the field ( EGA, FGA and SGA). I haven't read almost any content from any of these, only textbooks which are ultimately derived from these papers, and I only know some of what's contained in the textbooks. My advisor certainly does not know most of what is contained in Grothendieck's papers. And this is only the "foundations" of the field, let alone the thousands of problems that have been solved in AG. So, it absolutely seems like AI will be used to sort of fill in the "convex hull" generated by our current knowledge, and this is mostly what will happen first (by people unwilling to spend $10 million+ on compute to solve a hard problem like OpenAI did for Navier-Stokes). The question is, who is going to read the massive influx of papers that are being created and submitted to the arXiv? People are considering splitting the arXiv into two different sites to handle this. Yeah, this is one of the big incoming societal issues I think. There's likely going to be a huge division between people who actually know things and can read, interpret, and understand what an AI is doing to help guide it (if they use it for whatever reason) vs. people that sort of just use it to do what they can't and remain ignorant. Also in this second camp are probably those that are too easily fooled into believing what it's saying is wise/a prudent response because of some general pro-AI biases, or because the output is different than what they would have thought of, etc. My girlfriend is a software developer, and she already sees this. There are people not on her team that use AI heavily, but also clearly don't understand the goal they're trying to accomplish or the details of their situation as they use it. It's very frustrating to try to work with them. Recently there was a series of meetings, it's somewhat of a long story, but in the last one these people were still confused about very basic details of the project (and it is not at all a complicated project, it's just deciding the best way to allocate some group IPs). At the conclusion, they asked for information about the project details so that they could send it to Claude to make a short summary. She was like "...I can just summarize what we are doing for you...". A more relatable (but less specific) example is Suno, the AI music generator. Adam Neely has a great video on it. I just think about how many people are going to *avoid* learning an instrument or any music theory because they have access to an AI music generator that can generate whatever they want. It's unfortunate. Of course, there is also the huge crisis in education...

Update: It seems the open letter partially worked! https://www.businessinsider.com/openai-caltech-ai-math-hackathon-backlash-anthropic-2026-9

I am a math PhD student. Because of this, I thought I'd talk about the recent achievements of AI from the perspective of someone within the mathematics community. I was originally going to type some kind of essay, but I don't think I can do better than the open letter to the Mathathon signed by a large number of mathematicians: https://docs.google.com/document/d/1IL0b2oG2KvvSnxn_DuXsNxuHaCefDJ9hNY1k9BP3kKQ/view?tab=t.0 There are also good talks given by Terence Tao on this subject. Here is one at the ICM: And here is a talk intended for a more general audience: The basic idea, which applies to many fields, such as math and software engineering, is that the real bottleneck is human understanding. For math, that is essentially the entire point of the field. If you have a machine that can generate a proof of some result by cobbling together existing knowledge, especially if it's done in an inelegant way that is difficult for humans to understand, then human understanding is not advanced very much, and no new mathematics was developed. That will only happen if someone reads the proof, takes time to extract the methods that were used to solve it, communicates them to others, and passes along the knowledge to "the next generation." For software engineering, the bottleneck is not in writing code (once you're sufficiently experienced with this), it's in understanding the problem you're trying to write code to solve and understanding the current state of your codebase so you know what changes to make and how to design your software on a high level. Even if you use AI to write code for you, you have to read the code the AI has generated and understand what it does, which takes time. Now, this is assuming we actually care about human welfare, that we want to have the humans understand what the AIs are doing, and that we want to have humans "directing things". It seems like many of the people at top AI companies assume that humans are not going to be in charge of the future in the next decade, but they're all in a race to the bottom and so they can't help but destroy everything (from their POV). I know people at some software companies in the bay area where everyone is just using AI to write thousands of lines of code each day, and nobody can peer-review the code because nobody knows what's happening anymore. I also have concerns for the funding of math. If funding primarily comes from people with money who only care about math for new applications that can make them more money, and if AI can give them results with none of the human understanding (since they don't care about that), the field as a whole could dry up if there are no resources provided for it. One practical way this would be bad (if we care about humans knowing what is going on) is if we lose a huge number of people that understand math well, we lose a huge number of people that can understand what the AIs are doing when they are doing math. Personally, though, it would also be very sad to see the field die. I could say more, but maybe I'll just leave it at it seems the field of mathematics is going to have to undergo a huge paradigm shift, but nobody exactly knows how.

It's surprising that you find the men around you fall into this fairly strict dichotomy. I'm a guy, so I don't pay attention to these things as much, but it seems like most of the guys who have been around me in my life were fairly normal. I would even say that beta soyboys who are into spirituality were much rarer than Trump supporters.

Pxie wanted him to give her a large amount of money, so much that she didn't believe he would ever be willing to do it (something between "pay for her law school" or "a million dollars" iirc). It seems she decided to sue him partially because of this and partially because she didn't feel he was trying to make things right/taking it seriously enough in general.

Is Notebook LM really so good? My one experience with it was feeding it a paper in pure mathematics straight from the arXiv, and I got the vaguest overview that you could possibly imagine. Sure, it was an extremely technical paper and there are limits to what I should expect. Maybe the system can't even parse such a paper very well. At the same time... the podcast was basically just two people getting sidetracked constantly, and when they were on topic, they could only make up incredibly banal analogies. Imagine two people with ADHD adlibbing a podcast over a topic they have zero knowledge about. It was even more concerning because I first heard about Notebook LM from other graduate students (most of them in some kind of science field) who seemed to find it valuable. It is better with less technical data? They can speak meaningfully and productively in a focused way? Maybe I should try it again with excerpts from a novel to see how nuch better it is.

A good (and fairly long) video a friend sent me: It talks about the evolution of Facebook's destructiveness to our society over time (in particular, the troll farms and Russian disinformation). And not just our society! At some point, Zuckerberg rolled out Facebook to Myanmar and other third world countries. It ended up severely amplifying racial tensions which lead to an ethnic cleansing. Pretty concerning; I don't see these issues going away anytime soon. Or improving at all -- Elon Musk bought twitter and basically just turned it into an unregulated Russian bot farm where misinformation is allowed to run rampant, often even further spread by Elon himself.

Democracy requires some foundational prerequisites to be true in order to work effectively. People need to have accurate and high-quality information available to them, and they need to be able to make sense of this information in a high-quality way. Having massive amounts of pure garbage flooding all social media platforms undermines the first prerequisite. The fact that most people have a low capacity for high-quality sensemaking (for various reasons: education, tribalism also fueled by social media) undermines the second. Russia knows this, and this is why they pay people like Tim Pool and Dave Rubin to spread tons of garbage for no other purpose than to sow discord and undermine our democracy. The idea of free speech has become weaponized to allow massive amounts of disinformation and propaganda to spread. People, say Dave Rubin, will say things like "the marketplace of ideas," because people superficially love the idea of this. Then they use it to continue to spread tons of garbage everywhere. It's obviously hard, though, because you also can't just start banning certain types of speech. If you try to ban misinformation, more than half of the country freaks out. If you don't ban it, they continue to be poisoned and the quality of the country slowly crumbles. The only way out is some kind of cultural enlightenment, I think, but this doesn't seem like a real possibility either when people are arguing about whether or not hurricanes are made by the government. I believe this issue will only continue to get worse in the future. The only positive shift that comes to mind is that I have a vague memory of some fairly popular podcasters recommending people to heavily cut down on their social media intake or to stay off of it, but I don't think this is enough.

I was born male and took estrogen for 1 year. I stopped nearly 2 years ago. My sexual attraction to women never changed, I met my current girlfriend near the end of my time on estrogen. This happens to many people, there are a lot of trans woman lesbians. Also you can reverse it — at least, for males taking estrogen — depending on how long you were taking hormones and individual factors. My body went back to normal over a few months, except for the fact that I have some fairly mild gyno now 😔. (Unless you meant different hormones?)

I think it's really going to depend on how much time you want to put into learning math. I'm fairly confident that, with enough time, you could learn enough math such that the exam isn't very difficult for you. This may require relearning a lot of basic math from scratch. I say this as somebody with a publication in pure math. However, given that you hate math, this may not be the best path forward; Leo's path may be a better alternative. I think the cost of his approach is that any time you do encounter math in your degree, it's always going to be fairly confusing and painful. For a psych degree, you probably won't need to endure too much, so I think this is a valid tradeoff to make. I think my only remaining worry is that the "memorize and grind through" approach may end up being more work in the long run, but it's hard to say. Deeply learning a lot of math concepts takes a lot of time as well.

Well, in some sense, getting a full-time job to do some math-related work for an arbitrary company already feels like selling my life. I will avoid working overtime, though, considering I already don't want to work full-time.

This is probably what I'll end up doing. I do want to start some sort of business or independent venture of my own at some point, though. It's just hard to know 1) specifically what I should do, as well as 2) when to do it. I have an idea in my mind of creating youtube content. Perhaps I can do that on the side, even if I have a job. It probably makes the most sense at this point to work a job with good pay for awhile, and acquire a lot of capital/financial independence.

I guess that's the thing. At one point, I did have a grand passion. I spent all of my time for several years doing pure math and became pretty good at it as a result. Good enough to publish as an undergrad, at least. Now that I realize how much of my life would be eaten up by struggling to comprehend increasingly meaningless complex abstractions if I go into academia, I am less interested in it. Thus, the passion has collapsed, but there is nothing left in its wake and so I feel sort of directionless. Of course, I am still going to college so it's not as if I am actually directionless, yet. I just don't know what comes next and I have no grander vision like I used to. Maybe you're right. I think it's reasonably expensive though. I've slowly depleted my savings over the years of going to college because math at university is so difficult that I can only work 15 hours a week—just enough to pay my bills. I'll keep this in mind for the future, though.