TiB is always made up of three short segments followed by five “quick links”. Here, in chronological order, are my favourite segments from TiB in 2020:
1: GPT-2, chess and the limits of explainable AI
, section 2): Before there was GPT-3, there was GPT-2 and it was already raising questions about how we think about human intelligence
3: The troubling epistemology of coronavirus
, section 2): Why did the establishment get coronavirus wrong for so long?
4: How do we “disrupt government”?
, section 1): Discussion of Marc Andreessen’s “It’s Time to Build” and ways to improve the quality of public administration
5: Lessons from coronavirus: is Singapore *too* efficient?
, section 1) Looking at the efficiency vs resilience trade off through the lens of coronavirus (including one of my all-time favourite anecdotes)
6: TikTok, China and ideological conflict
, section 2) The geopolitics of technology was frequently in the news this year - and TikTok became one of the most prominent flashpoints
7: The tragedy of lost genius
, section 1) What’s the cost of failing to identify and nurture extreme talent? We discussed a fascinating paper
that suggests it’s huge
8: Can venture capital work outside startups?
, section 2) VC has lots of benefits in “extreme risk, extreme reward” domains; we discuss why it’s been hard to make work in philanthropy and science
9: Why the best ideas don’t always win: AI edition
, section 1) A discussion of Sara Hooker’s fascinating paper on “the hardware lottery
” and why science and technology can get stuck in dead ends
10: The rise and rise of technological sovereignty
, section 1) We review one of the big themes of the next decades - why nations increasingly want independent capabilities in emerging technologies