I choose to think of AI as being more akin to a calculator than a potential mathematician. As a tool, not an outsourced mind. As a brainstorming companion and source of potential hidden inspiration. In those regards, it is incredibly powerful and can amplify and accelerate the potential of an already curious, humble human mind. I feel blessed that I was raised and developed in a pre-AI academic environment as the rigors I encountered could only be confronted with my own faculties. I also had the knowledge that, from a competitive standpoint, my peers were on an equal footing. I do worry that today's true beautiful minds might be compelled to lean too heavily on AI if they see less dedicated, less talented peers achieving so-called accolades from their educators by using AI. I was never tempted with that. All told, though, if AI is viewed as a tool in a broader toolkit, it seems on balance to be a net positive. I doubt someone as thoughtful as Einstein would've outsourced the best parts of himself to AI if it were available to him, but I also highly doubt your thesis that he would've avoided it altogether.
I think this article identifies a real danger, but draws the wrong conclusion. It treats AI mainly as an answer machine. Used that way, it can absolutely weaken us. A tool becomes intellectually damaging when it removes the struggle through which we develop judgment. But AI can also be a thinking partner, helping us make connections and explore ideas we might not reach alone.
There’s some irony in the article dismantling the myth of Einstein as a lone genius. It describes how his thinking developed through the Olympia Academy, conversations with colleagues, wide-ranging reading and help from people with knowledge he lacked.
Einstein is often credited with saying, “My pencil and I are more clever than I am.” Whether or not those were his exact words, the idea feels right. Tools can extend our ability to think. AI is a far more powerful and dangerous version of that possibility. Einstein might have rejected outsourcing his thought, but I’m not convinced he would have rejected a tool that helped him think beyond what he could think alone.
Ethan — you diagnosed a real symptom but blamed the wrong thing.
A take-home exam an AI can ace isn't proof your students decayed — it's proof the test measured what's already automated. By your own Koyejo section, that's a broken benchmark: gamed at the surface, scoring agreement not correctness. You held the right tool and swapped it for moral outrage.
So the question isn't "resist AI." It's "why test people for what machines now do?" Let the machine do the machine-work; move the human bar up to what it can't reach. And the Einstein line you quote — "not the learning of facts, but training the mind to think what can't be learned from textbooks" — swap "textbooks" for "AI" and it's my argument, verbatim. Your own witness testifies for me.
You aimed at the student. The target was the test — and the training schedule built for a world that no longer exists
The Brown story doesn’t prove what you think it proves. When a take-home exam can be aced by AI, that isn’t evidence students have decayed — it’s evidence the exam was measuring something already automated. You spend half this piece on Koyejo’s point that a benchmark can be gamed by rewriting the surface while the substance stays hollow, that a score measures agreement, not correctness. A take-home test an AI can solve is exactly that broken benchmark. You diagnosed the disease and then used a symptom of it as moral proof.
So the real question isn’t "should we resist AI." It’s "why are we still testing for things AI now does?" If a task can be thought through by a machine, the humane move is to let it, and push assessment up to what machines still can’t do — posing the right question, spotting the anomaly, judging between several answers that are all internally consistent, and owning that judgment. That is not my rebellion against your essay. It’s your own witness: the Einstein line you quote — "the value of an education is not the learning of facts… but the training of the mind to think what cannot be learned from textbooks" — swap "textbooks" for "AI" and it is my argument, verbatim.
And there’s a floor you never reach. "Train once early, then work off that training for life" was built for a world where knowledge outlived the worker. When methods turn over every few years, that model is structurally broken — not morally, structurally. No amount of willpower fixes a schedule that guarantees obsolescence mid-career. You diagnosed a real symptom, but prescribed willpower for an institutional and structural problem. You’re not defending the human mind. You’re defending the schedule we used to train it — and mistaking one for the other.
Dittos for Nikola Tesla, who surpassed them all in scientific intellect and invention. Indeed, he was truly gifted by God, unlike the others. Otherwise, why would the federal government maintain secrecy of his most innovative scientific inventions.
The message in this post matters a great deal. The fundamental posture we take as humans, cultivating critical thinking and intellect so that we can grow an intellectual community and contribute to it, is exactly what gets laid bare now, in the age of AI, standing in front of a tool with more computational power than we have.
I share your sense of how much this matters. That is why I have been studying the decisive modes of thinking through which humans contributed to the historical progress of mathematics and physics, with the goal of establishing a new axis for the field of knowledge. I have named it ‘The Cycle Of Divergent And Preservation Thinking.’
I have put a link to the post below, if you would like to take a look.
Great article! And I agree we need to be very cautious about how we use AI. We’ve already suffered the Google Effect, eroding our ability to remember facts and figures. Now LLM’s risk displacing our basic reasoning skills 😬
Ai is great! But you have to use it properly. I set up rules that require it to look for peer-reviewed articles, provide links to those articles, make a brief statement as to how those the material in those links were used.
Now the speech to text function on my pixel 10 is pretty bad since they started using AI instead of rules.
I choose to think of AI as being more akin to a calculator than a potential mathematician. As a tool, not an outsourced mind. As a brainstorming companion and source of potential hidden inspiration. In those regards, it is incredibly powerful and can amplify and accelerate the potential of an already curious, humble human mind. I feel blessed that I was raised and developed in a pre-AI academic environment as the rigors I encountered could only be confronted with my own faculties. I also had the knowledge that, from a competitive standpoint, my peers were on an equal footing. I do worry that today's true beautiful minds might be compelled to lean too heavily on AI if they see less dedicated, less talented peers achieving so-called accolades from their educators by using AI. I was never tempted with that. All told, though, if AI is viewed as a tool in a broader toolkit, it seems on balance to be a net positive. I doubt someone as thoughtful as Einstein would've outsourced the best parts of himself to AI if it were available to him, but I also highly doubt your thesis that he would've avoided it altogether.
MetalTOMB...EXTREMELY well said
I think this article identifies a real danger, but draws the wrong conclusion. It treats AI mainly as an answer machine. Used that way, it can absolutely weaken us. A tool becomes intellectually damaging when it removes the struggle through which we develop judgment. But AI can also be a thinking partner, helping us make connections and explore ideas we might not reach alone.
There’s some irony in the article dismantling the myth of Einstein as a lone genius. It describes how his thinking developed through the Olympia Academy, conversations with colleagues, wide-ranging reading and help from people with knowledge he lacked.
Einstein is often credited with saying, “My pencil and I are more clever than I am.” Whether or not those were his exact words, the idea feels right. Tools can extend our ability to think. AI is a far more powerful and dangerous version of that possibility. Einstein might have rejected outsourcing his thought, but I’m not convinced he would have rejected a tool that helped him think beyond what he could think alone.
Ethan — you diagnosed a real symptom but blamed the wrong thing.
A take-home exam an AI can ace isn't proof your students decayed — it's proof the test measured what's already automated. By your own Koyejo section, that's a broken benchmark: gamed at the surface, scoring agreement not correctness. You held the right tool and swapped it for moral outrage.
So the question isn't "resist AI." It's "why test people for what machines now do?" Let the machine do the machine-work; move the human bar up to what it can't reach. And the Einstein line you quote — "not the learning of facts, but training the mind to think what can't be learned from textbooks" — swap "textbooks" for "AI" and it's my argument, verbatim. Your own witness testifies for me.
You aimed at the student. The target was the test — and the training schedule built for a world that no longer exists
The Brown story doesn’t prove what you think it proves. When a take-home exam can be aced by AI, that isn’t evidence students have decayed — it’s evidence the exam was measuring something already automated. You spend half this piece on Koyejo’s point that a benchmark can be gamed by rewriting the surface while the substance stays hollow, that a score measures agreement, not correctness. A take-home test an AI can solve is exactly that broken benchmark. You diagnosed the disease and then used a symptom of it as moral proof.
So the real question isn’t "should we resist AI." It’s "why are we still testing for things AI now does?" If a task can be thought through by a machine, the humane move is to let it, and push assessment up to what machines still can’t do — posing the right question, spotting the anomaly, judging between several answers that are all internally consistent, and owning that judgment. That is not my rebellion against your essay. It’s your own witness: the Einstein line you quote — "the value of an education is not the learning of facts… but the training of the mind to think what cannot be learned from textbooks" — swap "textbooks" for "AI" and it is my argument, verbatim.
And there’s a floor you never reach. "Train once early, then work off that training for life" was built for a world where knowledge outlived the worker. When methods turn over every few years, that model is structurally broken — not morally, structurally. No amount of willpower fixes a schedule that guarantees obsolescence mid-career. You diagnosed a real symptom, but prescribed willpower for an institutional and structural problem. You’re not defending the human mind. You’re defending the schedule we used to train it — and mistaking one for the other.
Dear Mr. Siegel, it is absolutely incredulous that you failed to mention either as a member of the discussion group or as a major contributor to his work; his wife Maric Mileva Please in the future do more complete research. https://www.encyclopedia.com/women/encyclopedias-almanacs-transcripts-and-maps/einstein-maric-mileva-1875-1948
I just checked it out Sarah.
Thanks for sharing this link.
I don't know if women were buried deliberately in the past or …
Dittos for Nikola Tesla, who surpassed them all in scientific intellect and invention. Indeed, he was truly gifted by God, unlike the others. Otherwise, why would the federal government maintain secrecy of his most innovative scientific inventions.
Un ejemplo en contexto de la manera de usar correctamente la IA y de los efectos de no usarla de la manera adecuada.
The message in this post matters a great deal. The fundamental posture we take as humans, cultivating critical thinking and intellect so that we can grow an intellectual community and contribute to it, is exactly what gets laid bare now, in the age of AI, standing in front of a tool with more computational power than we have.
I share your sense of how much this matters. That is why I have been studying the decisive modes of thinking through which humans contributed to the historical progress of mathematics and physics, with the goal of establishing a new axis for the field of knowledge. I have named it ‘The Cycle Of Divergent And Preservation Thinking.’
I have put a link to the post below, if you would like to take a look.
https://asymmetry0.substack.com/p/spin-off-why-divergent-thinking-is?r=8k16kp&utm_campaign=post&utm_medium=web
https://asymmetry0.substack.com/p/divergent-thinking-what-if-the-new-c58?r=8k16kp&utm_campaign=post&utm_medium=web
https://asymmetry0.substack.com/p/divergent-thinking-what-if-the-new-154?r=8k16kp&utm_campaign=post-expanded-share&utm_medium=web
https://asymmetry0.substack.com/p/divergent-thinking-what-if-the-new-295?r=8k16kp&utm_campaign=post&utm_medium=web
https://asymmetry0.substack.com/p/divergent-thinking-what-if-the-new-4e5?r=8k16kp&utm_campaign=post&utm_medium=web
Einstein was not stupid. He would have used all available resources. The curious mind does not eschew new ideas.
Would he have sat there scrolling social media feeds all day long, I think not.
It surprises me that this essay does not mention the mathematician and physicist Mileva Marić (Einstein's first wife) at all. Why the oversight?
Great article! And I agree we need to be very cautious about how we use AI. We’ve already suffered the Google Effect, eroding our ability to remember facts and figures. Now LLM’s risk displacing our basic reasoning skills 😬
Ai is great! But you have to use it properly. I set up rules that require it to look for peer-reviewed articles, provide links to those articles, make a brief statement as to how those the material in those links were used.
Now the speech to text function on my pixel 10 is pretty bad since they started using AI instead of rules.