My sense is that I have no idea what future college teaching, or future research, will look like, once systems like ChatGPT have matured. (I should have had, but strangely did not have, a similar sense of big unknown future developments when the internet, and mobile internet access, were invented.)

One thing that has surprised me is how confidently ChatGPT cites results from maths books which, when I look those books up, don't actually exist.

I wonder how NATO will take this.

Can you elaborate? In which way is the understanding of AI "shallow"?

The September 11 attacks in the US in 2001; the July 2005 attacks on the London underground which I experienced directly; the November 2015 attacks on Paris; the US wars in Afghanistan and Iraq; the rise of the caliphate in Syria and Iraq; it all seems a sequence of related tragedies.

Ambition or hubris?

Let me re-phrase: do you think it is pure coincidence that coin tosses tend to produce evenly distributed results, or is it the result of some objective reality, and we can usefully describe that reality using probabilities?

(Creamer-McLean is not a non-existence result.) The fact that two modeling approaches both imply counterintuitive results does not tempt me into thinking that I have to affiliate myself with one of them.

If probabilities are purely subjective, then no auction is expected revenue maximizing with risk neutral bidders. The mechanism designer can improve upon any auction by further exploiting differences in bidders' subjective probabilities.

I want to assail the fortress :) Is it pure coincidence that people's probabilities for some events are closely aligned, whereas for other events people differ widely? Relatedly: If probabilities are purely subjective, wouldn't we expect to see more betting than we do in practice?

Why not relitigate the common prior assumption? Isn’t it fun?

Let a hundred flowers bloom.

My sense is that having strong opinions on notation is a good way to make one’s life as a theorist unhappy.