[mitchellberry]

This article seems more higher education or tax policy discussion rather than about AI/ML.

Still intrigued though, is "world rankings" possible to determine or even valid in a multinational corp world?

What was the last actual eureka moment in ML?

Apart from some hello worlds know little on the subject, from the outside looks like the heavy work was done long ago and we've only recently had the hardware advances to make it a reality. Seems to be incremental improvements on models and a myriad of new ways to implement neural nets, but the real improvements are coming from nvidia/amd engineers making gpu advances.

Hardware has moved on but we still in the same position of engineering tradeoffs and physical limitations. Undoubtedly there's wizards out there who can work wonders, and there's people less adept with nation-state resources who could can achieve the same result.

I'm partial to the opinion that being a world leader in ML right now has more to do with your hardware and budget rather than individual talent.