Great video. I met Leslie once, sat on the bus beside him on the way to a conference around 8 years ago. He wasn't the chattiest, but you bring up his work, he likes to talk. I think he was just over 70 years old, but still incredibly sharp. At the time Microsoft Research were shutting down their valley office, but they would still let him come in there - last one to put the lights out (metaphorically for computer science research at the big IT companies). Nowadays, he couldn't do the research work he did there and at other places at any big IT company - it's R&D, with the emphasis on "D".
I believe VMWare "adopted" Microsoft's research team, but that's the last I heard of the team. These days the most interesting corporate research happens at Google, Nvidia, OpenAI. I guess the forefront of research has moved onto ML and many old school researchers got left behind.
There's lots of other research happening all over, but gets little attention probably due to non-existent or otherwise poor marketing beyond publishing papers.
> Nowadays, he couldn't do the research work he did there and at other places at any big IT company - it's R&D, with the emphasis on "D".
I, on the other hand, am amazed by the progress being done today, and the research paper released on an almost weekly basis by companies like Google, Microsoft and co. Just look at the language and image models that are released, the progress on computer vision etc.
I played with the playground of OpenAI GPT-3 and I am blown away at the result. It's not perfect, but it's orders of magnitude better than what we had easily access to just few month ago. I have started using it in my day-to-day life on some specific tasks, and I can't wait to see the next versions. I can't even start to fathom the amazing products people are going to build around it in the next decade.
It's absolutely true that these models/research serve the purpose of the company building them... but even Leslie says in the video that he did his entire career in the industry because that's where he saw many interesting challenges, and that his whole research was a means to an end.
R&D is still alive and kicking, and is going faster than ever. You might just be looking at the past with rose tainted glasses, and needlessly undermining the present.
The problem is the usage of the label 'AI'. There are basically 4 uses of the the label: 'General AI' (popular among layfolk), 'stuff computers can't do yet' (popular among developers), 'stuff we just figured out how to do' (popular among sales people), and '2nd order solutions' (popular among academics).
