Instead of writing that the results in this paper have been obtained using AI, we could write that the results in this paper could not be proved by AI. This can be used as a proof of non-triviality or novelty that could facilitate the evaluation of the paper. Practically, one can share a conversation with the model.

This goes back to something that I’ve always been interested in: how to define banality. My definition was in terms of kolmogorov complexity, so something is banal if it has low kolmogorov complexity given all the rest that is out there. Here, I’m also referring to novels, movies, music, etc., not just math. Naturally this definition is impractical, and it is interesting that large language models can give a practical definition of something similar.

In some communities, compressors like ZIP are used as a proxy for Kolmogorov complexity. It would be interesting to try to use large language models instead or in combination with compressors.

There is a lot of anxiety about ai wiping off mathematics, including theoretical computer science. It’s funny that we wanted ai to cut *their* jobs, and instead it’s *our* jobs that are cut (maybe). My expectation of what is going to happen is rather flat, and I am open to various scenarios. Still I wanted to make some points.

First, at the moment of this writing, I am not so worried about ai killing the field. There are so many problems in math, and the literature is so unmanageably vast and technical, that I am not particularly shocked that using massive resources one can solve *some* problems. It is very different if the resources can solve *the* problems. For example I, and I am sure many others, have tried to use ai to solve problems in computational complexity and so far didn’t get much. I do find ai to be a very useful assistant, but so are many other things. It may be that the next level of solving target problems (as opposed to finding targets) may prove the most difficult to reach.

I want to suggest an option for the community to put ourselves in a position of strength, in case one really fears the impact of ai. I think ai can easily enough be “frozen” and made much less useful for future research. The way to do this is simple: We could stop feeding it. It was humiliating enough to post papers online only to be asked later by the publisher to pay for “gold open access.” But now that there is this new way to exploit, plagiarize, and monetize our creations on a massive scale, it may be too much. Suppose starting immediately all new math is communicated in ways that ai can’t easily scrape. There are many ways to do this; we could still put papers online, but allow only much more limited access, compatible with human beings but not ai scraping. It coud be similar to what is done for example at the Internet archive, where you can read a book but not easily download it. I am not going to go more in details. While ai would remain very useful for things on the table until that moment, I think it would quickly become much less useful for new lines of research, series of papers building on each other, etc. This would put the community in a position of strength as keeper of knowledge. After a while, things could be reassessed.

We should not forget that the models can do math only because back then we chose to be nice and so taught them how to do it for free.

Enrollment in computer science is declining, out of fear that AI will wipe out much of the demand for software engineers. So here are the jobs of the future:

• Real estate agent. Indeed, it will be pretty hard for AI to steal their added value: zero.
• Driver. We will always need a human to shuttle people to airports and drive food trucks coast to coast, obviously.
• Banker. Nothing can beat the personal touch of a slender banker, all smiles and a nice suit, meeting you in a brick-and-mortar branch to tell you which buttons to push on your phone — the service courtesy of your account fees.
• Electrician. So you can clean the vents in the data centers. If you are lucky, you get to install solar panels.

In 2008, when I was on the job market, computer science was at its nadir. The people behind the doors I was knocking on told me kids were told to study biology and law instead. What happened instead is that people entering the field precisely at that point were going to graduate at a very good time.