Isn't this line of thinking which has held physics back for the past few decades? I thought that physics is gradually moving to being more data driven than being driven by ideas of beauty. Sabine Hossenfelder has written quite a bit about this.
The problem with theoretical physics, in my external perception, is more political, within academia.
- selecting for the H index, that is, what will be most cited, leading to strong convergence around "consensual" or "popular" research, at a huge detriment to diversity of ideas
- one negative effect is a major push for incremental publication (equivalent of a LOC KPI for coders... it's just wrong). “What did you publish this year? How many times was your name in a paper somewhere?”
- putting funds and institutional support behind people you like as opposed to people with the most promising ideas. That's pretty much the story of string theory, and see how despite the failure they are still "winning" politically, occupying top positions and feeding the mainstream vulgarization, whereas those who "lost" decades ago are still AWOL even as their ideas are freshened up by a new generation, whose future against the institutional pyramid is all but guaranteed.
We might still be there (a situation of total stagnation, since the 1970s, at the theoretical level) by 2040 if something doesn't change profoundly in the way university politics have such a compelling handle on research, thus on our progress in science in general.
My 2cts but I'm parroting, people who now speak up in academia.
What I do know about, ML, is all but dead currently in MIT, Stanford, etc. It's just monkey coding applying whatever the industry likes for short-term benefits (again this "incremental"-only approach, playing it safe with 0.001% gains that are, sure, predictably doable). There's no actual research on "AI" let alone the general notion of "intelligence". Not even in neuro. I fear we've lost sight of the importance of basic research, as basic as it gets in a given field, because industry deals run the show in terms of funding. Yet look what Bell Labs or MIT did in the 1970-80s, surfing on the last wave of fundamental breakthroughs of the 50-60s. These days are long gone.
