Nowadays AI can solve problems even better than mathematicians, so is there a reason for me to still pursue a PhD? The academic skill gained from a PhD would become obsolete by the next few years.
Eventually there won’t be such job as a mathematician, as AI can effectively (i.e. not always, but usually) act as the Hilbert’s dream oracle: send a statement, and the AI decides whether it’s true or false, and gives the proof (or disproof). The only job of human is to retrieve the proof and try to understand it.
In the artistic sense, this is sad, and human endeavor in mathematics is getting to its extinction. In the scientific sense, this is hopeful, since we have better tools at finding truth!
You should probably not expect to get a job writing certain kinds of proofs, but I’d expect the bigger part of most PhDs to be training judgement and taste (what matters, what approaches make sense), which are the two things AI is worst at.
For context, none of the software engineers I work with write meaningful amounts of code anymore, but we still have tons of work: deciding what to do, high-level design, code review, coordination, and ensuring rollouts are done safely.
Who knows what AI will look like in 4-6 years though.
Nowadays AI can solve problems even better than mathematicians, so is there a reason for me to still pursue a PhD? The academic skill gained from a PhD would become obsolete by the next few years.
Eventually there won’t be such job as a mathematician, as AI can effectively (i.e. not always, but usually) act as the Hilbert’s dream oracle: send a statement, and the AI decides whether it’s true or false, and gives the proof (or disproof). The only job of human is to retrieve the proof and try to understand it.
In the artistic sense, this is sad, and human endeavor in mathematics is getting to its extinction. In the scientific sense, this is hopeful, since we have better tools at finding truth!
So, it is what it is, and that’s life, I guess...
You should probably not expect to get a job writing certain kinds of proofs, but I’d expect the bigger part of most PhDs to be training judgement and taste (what matters, what approaches make sense), which are the two things AI is worst at.
For context, none of the software engineers I work with write meaningful amounts of code anymore, but we still have tons of work: deciding what to do, high-level design, code review, coordination, and ensuring rollouts are done safely.
Who knows what AI will look like in 4-6 years though.