[Gatsky]

It's easy to get upset about all this. Things happen in cycles though. Academia is currently in a lull due to poor incentives, oversupply of research staff arising from the massive growth of universities and the democratisation of higher education, and the general devaluation of experts and technical skills. I also see that in certain foundational areas privately funded research is outperforming academia, sometimes by silly margins (eg Deepmind's protein folding work, Google and quantum computing). The protein folding is a good example of why - Deepmind had something like 10+ senior researchers working on the project, whereas academia has overworked flustered PIs who do admin mainly and a few students starting from scratch each year. This discrepancy can only exist because academia doesn't really care a lot about actual progress. Another interesting phenomenon lately has been the total deluge of deep learning papers. Once the cat was out of the bag, there was a literal explosion of research output in a very short time. One can only conclude that there are a lot academics sitting around looking for the next big thing (rather than working to create it). This is a bit cynical perhaps. There is still progress being made, but as others have noted, it definitely seems to be slowing overall.

I think actually we haven't seen the worst of it. My theory is that the rapidly ageing demographics of the world, which it is important to note is totally unprecedented, will have profound impacts on research and life in general. Mostly, it will be less interest in and funding for research. The cost of caring for the elderly is part of it (see where government revenue goes in more socialist countries, or observe the huge pension liabilities coupled with increasing life span in the USA and some of Europe). But there could be less tangible factors, like the willingness of a more elderly society to support research which will only bear fruit long after many of them are dead. This is understandable. Society will fund as much research as it sees fit. It is hard to make an actual moral argument for funding the kind of extremely expensive, highly technical and incremental pursuit that science has become.

[cycomanic]

I want to just add that most of the current "leading research in ML" at Google et al would not be possible without the years of foundational research conducted primarily at universities before the recent (3rd?) ML hype. Also pretty much all the Google researchers were hired away from universities I believe. That is the thing, industry hardly ever does high risk research that might only pay dividends in a decade or more.