Ian Hogarth and Nathan Benaich published their annual State of AI report
. It’s a real tour de force
- in my view, the best overview of what matters and what’s changing in the world of artificial intelligence. Lots of themes will be familiar to Thoughts in Between
readers, including GPT-3 and language models, Clearview and regulation of facial recognition, the battle for AI talent, the geopolitics of semiconductors, and more (It’s also nice to see so many Entrepreneur First
companies featured, including Tractable, CloudNC, PolyAI and Papers with Code)
Ian and Nathan note a growing tension between academia and industry over talent. Big Tech AI labs have poached dozens of professors from leading universities: “no AI professor left in 2004, whereas 41 left in 2018 alone”. Given the extraordinary remuneration available to machine learning PhDs in the private sector (see e.g. TiB 11 and TiB 71
) this is perhaps inevitable, but is also a long term risk for the sector. And, as the report notes, it affects entrepreneurship too: this paper
suggests that when professors leave academia, fewer of their students start startups.
One thing I love about this report is the willingness to make public predictions. Their 2020 predictions were pretty good - 4.5 out of 6 - so it’s worth a look at those for 2021 on slide 172. Two touch on the theme of technological sovereignty, which we discussed last week in TiB 134
. First, they think NVIDIA’s proposed acquisition of Arm will be blocked. Second, they predict a surge of Chinese and European investment into homegrown AI defence startups, given the US’s plays in the space. I’m more confident on the second than the first, but both are important spaces to watch.