Following Terry Tao's blog he seems to be engaging in forced self-directed cognitive behaviorial therapy, he is trying to come to terms with loss while developing a new framework for himself and other as to where the value is. It is cool to see. It is better to engage than to be in denial and just fall into reactionary behavior.
I do not think I have such a well-developed concept of myself in relationship to my field as he does, I just found it hard to maintain and its value wasn't sufficient to invest that effort. I think the difference is likely that I am a "pragmatic engineering" type who does "the next right thing" and I do not need an overly complex over-arching concept of myself in relation to my craft.
This is entirely ahistorical and a typical remote psychoanalysis.
Tao has been pushing for AI and Lean for a long time and didn't like the way OpenAI handled the Oct. 6th release. He has now been pushing for AI owned by universities.
That is the public way he navigates the situation, who knows what his real game is.
He is one of the people that brought about this situation.
In any case: "reactionary"? Are we using Soviet language to single out perceived dissidents?
> This is entirely ahistorical and a typical remote psychoanalysis.
I've been reading the posts on his blog (and I should be clear, many posts are guest posts, so it is a collective cognitive behavior therapy session):
Also did you notice he never posted more than 10 posts per month until September? These are definitely posts trying to come to terms with the new reality and searching for value in a field where humans are not pre-eminent:
I also see this post that that mathematics will be needed as interpreters of complex AI generated proofs as a form of denial in that this is just a current limitation of AI, and will likely be fixed soon:
> In any case: "reactionary"? Are we using Soviet language to single out perceived dissidents?
When new technology is introduced that is disruptive a lot of people fall into "reactionary" stances.
The first definition of reactionary that I found was this:
"A reactionary is a person or policy that opposes political, social, or economic change and seeks to reverse progress to restore an earlier, idealized state of society."
The systems need Lean, training material, benchmarks, data labeling, experts, publicity and more. OpenAI gave out free researcher accounts to collect more training material.
All of which was provided and encouraged by a small group like Tao, Gowers, Tsimerman, Avigad, Schmitt etc. It isn't just "an algorithm".
Human pursuits cannot be separated from the humans who pursuit it. In the pursuit of knowledge, what ultimately drives people is what they themselves can understand, and the meaning these understanding provides, both based on their own experience and people whose opinions they value. Humans do not pursuit Platonic knowledge for their own sake, and those who claim to do it, often are satisfying their egos in a different way.
> Human pursuits cannot be separated from the humans who pursuit it.
But it's pretty clear you can chuck "human pursuits" in the garbage, in favor of what really matters: intensification of the market system and wealth accumulation for its own sake.
> In the pursuit of knowledge, what ultimately drives people is what they themselves can understand, and the meaning these understanding provides, both based on their own experience and people whose opinions they value.
It's pretty clear the message we're getting is "what [people] themselves can understand, and the meaning these understanding provides" doesn't matter, and if you care about that you're an sentimental piece of garbage.
This is one of the best posts I've seen on this topic in a while.
I finished a math PhD before I decided to get out of academia and into startups. Since then, the older I get the more of a Platonist I become, especially now that I've taught my daughters a bunch of math (I'm their homeschool math+science teacher). And the more of a Platonist I become, the less I care about who made a particular discovery. Who cares?! It seems obvious (and quite beautiful) to me now that the same discoveries must be getting repeated again and again across the universe by alien mathematicians, and eventually by their AIs.
You can still discover it for yourself even if AI discovered it first (or some other human, or an alien on another planet). It is fun to figure out how something works, in and of itself
Absolutely. One of the best pieces of advice my PhD advisor gave me was that to really understand a particular result I would usually have to redo the proof of it myself (even after reading someone else's proof).
From the outside seeing academics wash over concerns and debates about this it's helped me sort of get more comfortable with around my own questions about what I want from math and computers. What I've come to realize it's that I just want tools that do a job and enjoy learning them. I also get gratification of being creative with them. I'm not in it for expanding human knowledge or seeing the beauty in the universe or the pleasure of understanding something difficult.
Watching Star Trek TNG as a kid, I vividly recall hearing Geordi or Data often say things like "run a level 3 diagnostic on the Worf Barrels" or whatever to the computer and wonder what the heck that would even be. I wondered where are all ship devops and programmers were hidden. Though I recall the one accurate part was Barclay in the Holodeck spending most of the time setting up his IDE.
Now I think I kind of have an answer. Maybe the level of diagnostic refers to the LLM effort level. Great. So then what are all the people on the Enterprise actually doing? My guess is they're all just being people and doing people stuff like exploring the galaxy and smugfacing the Romulans. It would be easier to just create a race of Datas to do it safer. The Datas could probably be Bluetoothed to the Ship's Computer for even more optimization.
I think some of the struggle is that as we summit each subsequent hill of challenge, the sunset (or is it a sunrise?) is becoming inescapably bright. We're faced with that big ol' "what's the point of all this?" question. What's the point of studying math? What's the point of exploring the galaxy? What's the point of standing under a stack of blue barrels?
I'm not a math guy. I dunno if a Fields medal comes with free Taco Bell for life or something else that would truly move me. But I love that I get to read books and learn cool things and feel that spark of joy when my understanding expands, even if the ending is right there with nothing stopping me from skipping to it.
> Watching Star Trek TNG as a kid, I vividly recall hearing Geordi or Data often say things like "run a level 3 diagnostic on the Worf Barrels" or whatever to the computer and wonder what the heck that would even be....
> Now I think I kind of have an answer. Maybe the level of diagnostic refers to the LLM effort level....
The real answer is that it was gibberish. IIRC, the writers would insert "[tech]" in the scripts, to be later fleshed out into technobabble.
I think it's the third, because ironically the Freudian one seems to be invented by Freud out of a narcissistic desire to compare the impact of his work to Copernicus and Darwin.
Don't underestimate the impact Freud had.. Even though he is now often seen as a pseudo-scientic Quacksalber, he opened up a world with psychoanalysis. He was an immense popularizer of clinical psychology, the subconscious mind (!) and mental illnesses.
He really showed that we humans are driven be desires more than anything else.
From the jump the premise seems reasonable. Having a too much of an ego is a source of all sorts of self-imposed problems or create roadblocks in interacting with others. But this literally brings up saintlike detachment as "not far from the truth" in describing the extent they're advocating for.
Feel like it's hard to accept egoless style, as in academia there is little but the glory to be had in the pursuits. In the publish/perish era, money is already stretched thin in the average research budget.
I want to believe, but research has been tightrope walking with no safety net for awhile.
So, either researchers get SF salaries? or? (first one is not gonna happen)
> we gratefully acknowledge the assistance of GPT Astra 6.0 and GPT 6.1 Sol in drafting this post.
Holy. This is the closest thing I've seen one of these AI guys get to accepting it as the machine god. "Dear god, we gratefully accept this divine gift of sustenance that we are about to consume."
That might be a good time to consider what the fuck is going on.
> That might be a good time to consider what the fuck is going on.
Marketing for OpenAI. When it comes from one of the best-known mathematicians, it must be quite efficient.
btw,
> This blog post was initially written in a different file format and converted using AI.
Why even mention this? And you don't really need an LLM to convert between file formats, there are dedicated tools that work well for this. Especially for a text that's essentially unformatted, a copy paste and a few searches & replaces will probably be faster.
Acknowledgement: a few fingers were used to type this comment. Possibly a couple of organic neurons. No AI whatsoever, except maybe in cited text.
The top AI usage is for converting into a different format (really like pdf2html or something), which is unrelated to AI slop and done by Tao himself and not the guest authors.
Following Terry Tao's blog he seems to be engaging in forced self-directed cognitive behaviorial therapy, he is trying to come to terms with loss while developing a new framework for himself and other as to where the value is. It is cool to see. It is better to engage than to be in denial and just fall into reactionary behavior.
I do not think I have such a well-developed concept of myself in relationship to my field as he does, I just found it hard to maintain and its value wasn't sufficient to invest that effort. I think the difference is likely that I am a "pragmatic engineering" type who does "the next right thing" and I do not need an overly complex over-arching concept of myself in relation to my craft.
Like many other posts on Tao’s blog, this is a post by guest authors, not by Tao himself.
This is a guest post. Tao has been quite consistent in his stance.
This is entirely ahistorical and a typical remote psychoanalysis.
Tao has been pushing for AI and Lean for a long time and didn't like the way OpenAI handled the Oct. 6th release. He has now been pushing for AI owned by universities.
That is the public way he navigates the situation, who knows what his real game is.
He is one of the people that brought about this situation.
In any case: "reactionary"? Are we using Soviet language to single out perceived dissidents?
> This is entirely ahistorical and a typical remote psychoanalysis.
I've been reading the posts on his blog (and I should be clear, many posts are guest posts, so it is a collective cognitive behavior therapy session):
Also did you notice he never posted more than 10 posts per month until September? These are definitely posts trying to come to terms with the new reality and searching for value in a field where humans are not pre-eminent:
https://terrytao.wordpress.com/2026/09/26/a-value-based-appr...
https://terrytao.wordpress.com/2026/09/19/why-do-we-need-hum...
https://terrytao.wordpress.com/2026/09/12/after-math/
I also see this post that that mathematics will be needed as interpreters of complex AI generated proofs as a form of denial in that this is just a current limitation of AI, and will likely be fixed soon:
https://terrytao.wordpress.com/2026/09/18/if-math-is-more-th...
> In any case: "reactionary"? Are we using Soviet language to single out perceived dissidents?
When new technology is introduced that is disruptive a lot of people fall into "reactionary" stances.
The first definition of reactionary that I found was this:
"A reactionary is a person or policy that opposes political, social, or economic change and seeks to reverse progress to restore an earlier, idealized state of society."
I think that fits with my usage.
What situation? The development of an algorithm that can prove mathematical theorems?
The systems need Lean, training material, benchmarks, data labeling, experts, publicity and more. OpenAI gave out free researcher accounts to collect more training material.
All of which was provided and encouraged by a small group like Tao, Gowers, Tsimerman, Avigad, Schmitt etc. It isn't just "an algorithm".
Human pursuits cannot be separated from the humans who pursuit it. In the pursuit of knowledge, what ultimately drives people is what they themselves can understand, and the meaning these understanding provides, both based on their own experience and people whose opinions they value. Humans do not pursuit Platonic knowledge for their own sake, and those who claim to do it, often are satisfying their egos in a different way.
> Human pursuits cannot be separated from the humans who pursuit it.
But it's pretty clear you can chuck "human pursuits" in the garbage, in favor of what really matters: intensification of the market system and wealth accumulation for its own sake.
> In the pursuit of knowledge, what ultimately drives people is what they themselves can understand, and the meaning these understanding provides, both based on their own experience and people whose opinions they value.
It's pretty clear the message we're getting is "what [people] themselves can understand, and the meaning these understanding provides" doesn't matter, and if you care about that you're an sentimental piece of garbage.
This is one of the best posts I've seen on this topic in a while.
I finished a math PhD before I decided to get out of academia and into startups. Since then, the older I get the more of a Platonist I become, especially now that I've taught my daughters a bunch of math (I'm their homeschool math+science teacher). And the more of a Platonist I become, the less I care about who made a particular discovery. Who cares?! It seems obvious (and quite beautiful) to me now that the same discoveries must be getting repeated again and again across the universe by alien mathematicians, and eventually by their AIs.
I wish more people read Plato's work, or did shrooms. Both probably works.
yeah but it's fun to discover things with your own brain
You can still discover it for yourself even if AI discovered it first (or some other human, or an alien on another planet). It is fun to figure out how something works, in and of itself
Absolutely. One of the best pieces of advice my PhD advisor gave me was that to really understand a particular result I would usually have to redo the proof of it myself (even after reading someone else's proof).
sure, and what you said is also not mutually exclusive to what OP said
So why should your children bother to learn math? Just have them prompt their AIs.
This essay seems to say, "people don't matter; all that matters is truth." And that all the policy decisions should follow from that assumed value.
Yikes. I fear for what mathematics will become if that becomes the majority viewpoint.
The future of mathematics is outlined in this amusing quasi-satire (satire tends to become reality in Silicon Valley):
https://proofsandprompts.com/2026/10/31/lean-me-softly-a-sat...
That is literally what some tenured people are pushing for.
From the outside seeing academics wash over concerns and debates about this it's helped me sort of get more comfortable with around my own questions about what I want from math and computers. What I've come to realize it's that I just want tools that do a job and enjoy learning them. I also get gratification of being creative with them. I'm not in it for expanding human knowledge or seeing the beauty in the universe or the pleasure of understanding something difficult.
Watching Star Trek TNG as a kid, I vividly recall hearing Geordi or Data often say things like "run a level 3 diagnostic on the Worf Barrels" or whatever to the computer and wonder what the heck that would even be. I wondered where are all ship devops and programmers were hidden. Though I recall the one accurate part was Barclay in the Holodeck spending most of the time setting up his IDE.
Now I think I kind of have an answer. Maybe the level of diagnostic refers to the LLM effort level. Great. So then what are all the people on the Enterprise actually doing? My guess is they're all just being people and doing people stuff like exploring the galaxy and smugfacing the Romulans. It would be easier to just create a race of Datas to do it safer. The Datas could probably be Bluetoothed to the Ship's Computer for even more optimization.
I think some of the struggle is that as we summit each subsequent hill of challenge, the sunset (or is it a sunrise?) is becoming inescapably bright. We're faced with that big ol' "what's the point of all this?" question. What's the point of studying math? What's the point of exploring the galaxy? What's the point of standing under a stack of blue barrels?
I'm not a math guy. I dunno if a Fields medal comes with free Taco Bell for life or something else that would truly move me. But I love that I get to read books and learn cool things and feel that spark of joy when my understanding expands, even if the ending is right there with nothing stopping me from skipping to it.
> Watching Star Trek TNG as a kid, I vividly recall hearing Geordi or Data often say things like "run a level 3 diagnostic on the Worf Barrels" or whatever to the computer and wonder what the heck that would even be....
> Now I think I kind of have an answer. Maybe the level of diagnostic refers to the LLM effort level....
The real answer is that it was gibberish. IIRC, the writers would insert "[tech]" in the scripts, to be later fleshed out into technobabble.
AI inflicted the 4th narcisstic wound on humanity, the other three being Copernican, Darwinian and Freudian, see https://en.wikipedia.org/wiki/Narcissistic_injury for the background.
I think it's the third, because ironically the Freudian one seems to be invented by Freud out of a narcissistic desire to compare the impact of his work to Copernicus and Darwin.
Don't underestimate the impact Freud had.. Even though he is now often seen as a pseudo-scientic Quacksalber, he opened up a world with psychoanalysis. He was an immense popularizer of clinical psychology, the subconscious mind (!) and mental illnesses. He really showed that we humans are driven be desires more than anything else.
From the jump the premise seems reasonable. Having a too much of an ego is a source of all sorts of self-imposed problems or create roadblocks in interacting with others. But this literally brings up saintlike detachment as "not far from the truth" in describing the extent they're advocating for.
I'm thinking maths are kinda fucked rn.
Feel like it's hard to accept egoless style, as in academia there is little but the glory to be had in the pursuits. In the publish/perish era, money is already stretched thin in the average research budget.
I want to believe, but research has been tightrope walking with no safety net for awhile.
So, either researchers get SF salaries? or? (first one is not gonna happen)
> we gratefully acknowledge the assistance of GPT Astra 6.0 and GPT 6.1 Sol in drafting this post.
Holy. This is the closest thing I've seen one of these AI guys get to accepting it as the machine god. "Dear god, we gratefully accept this divine gift of sustenance that we are about to consume."
That might be a good time to consider what the fuck is going on.
> That might be a good time to consider what the fuck is going on.
Marketing for OpenAI. When it comes from one of the best-known mathematicians, it must be quite efficient.
btw,
> This blog post was initially written in a different file format and converted using AI.
Why even mention this? And you don't really need an LLM to convert between file formats, there are dedicated tools that work well for this. Especially for a text that's essentially unformatted, a copy paste and a few searches & replaces will probably be faster.
Acknowledgement: a few fingers were used to type this comment. Possibly a couple of organic neurons. No AI whatsoever, except maybe in cited text.
Crazy to see AI slop being posted on tao's blog.
Isn't it strange how the AI disclaimer is at the end of the writing instead of the start. Almost as if they are trying to hide it.
You must have some variant of banner blindness, because another mention of AI use is also at the start of the text.
The top AI usage is for converting into a different format (really like pdf2html or something), which is unrelated to AI slop and done by Tao himself and not the guest authors.