View profile

TiB 155: National AI strategies; scale and ethics; compressing the economic sphere; and more...

Revue
 
 
March 9 · Issue #155 · View online
Matt's Thoughts In Between
This week: A look at the 2021 “AI Index” report; thinking about scale and ethics in tech with Sarah Drinkwater; the challenges of “shrinking the economic sphere” and more…

Welcome new readers! Forwarded this email? Subscribe here. Enjoy this email? Forward it to a friend.
The rise and rise of national AI strategies
The 2021 AI Index Report was published this week and is really excellent. You can see a Twitter-length summary of the highlights, but I recommend browsing the whole thing. It presents extremely rich data across a huge range of topics, including technical progress, AI education, the economy and more. The overwhelming impression is of a discipline that continues to accelerate: almost every chart is “up and to the right”.
Two policy-related areas that we’ve discussed before are worth particular attention. First, the US’s attractiveness to international machine learning talent continues to grow. Almost two thirds of new PhD candidates in AI studying at US universities are international students, almost twice the proportion of a decade ago. Being the destination of choice for global talent is a huge strategic advantage (and an area where the UK, as I suggested in TiB 149, still has the opportunity to be a global winner).
Second, the section on global AI policy is impressively comprehensive: the report lists 32 countries with national AI strategies so far, with 22 others in development. But, as I’ve noted before, what’s striking is how (relatively) small the sums being invested in this space are. The US DOD reports spending $1bn on AI R&D (about half of the US Government total), which is roughly comparable in magnitude to Google’s DeepMind, which spent around $1bn in 2019. That said, the shape of the curve is striking: we’re clearly at the very beginning of the era of government interest in AI.
If US government AI spend were a startup, it could SPAC tomorrow
If US government AI spend were a startup, it could SPAC tomorrow
Responsibility, ethics and scale
The latest TiB podcast episode is a conversation with Sarah Drinkwater, Director of Responsible Tech at Omidyar Network. I highly recommend Sarah’s writing about responsible tech and what it means; we cover a lot of these topics in the conversation. We discuss some of the thorniest topics at the intersection of tech and society, such as how to regulate Big Tech, how to build positive communities online and the role of talent in influencing tech companies’ behaviour (a favourite TiB theme - see TiB 80 and TiB 149).
One of the most interesting segments was our discussion about the impact of scale, which starts around 16 mins in. A lot of the ways we think about responsibility and ethics break down at very large scale (We talked about this in the context of the work of philosopher Shannon Vallor way back in TiB 33). As we discussed in TiB 111, one of the attractions of Milton Friedman’s idea that the “social responsibility of business is to increase profits” is that it scales… but it also creates a situation where we have to worry about Big Tech’s foreign policy.
Sarah has done excellent work on how to nudge tech companies to be more responsible, but what if it’s the least responsible companies that grow fastest and so have the biggest societal impact? As my friend Alex argues, this would explain why ex post it often looks as though tech companies are “Bad” (though it’s worth pondering why this seems true, in Alex’s terms, for a Facebook or an Uber, but not a Stripe or a Hopin). Scalability is an essential (and welcome!) fact of company building in the internet era, but there’s no escaping the societal challenge of managing its consequences.
Should we compress the economic sphere?
John Cochrane has a good and provocative post on Europe’s terrible economic performance (Thanks Marc for the link). Over the last 20 years, Europe has lost ground to the US on a GDP per capita basis: the US is more than 50% better off than the UK and 96%(!) better off than Italy. You could argue this doesn’t matter so much: GDP isn’t everything and it’s legitimate to opt not to maximise it. You could point to most of Europe’s choice to work shorter hours than the US, for example (though the difference doesn’t explain the size of the gap).
Lots of smart people are gravitating to this idea. This excellent conversation between Yancey Strickler and venture capitalist Albert Wenger is worth a read (Thanks Ian for the link). Wenger, echoing ideas found in both Marx and Keynes, talks about the opportunity to “compress the economic sphere” and put more of our attention into non-economic activity. (Wenger’s expands on these themes in his World After Capital, which I recommend). Nevertheless, I’m not sure about this, for two reasons.
First, I suspect there is a sort of geopolitical prisoner’s dilemma dynamic to compressing the economic sphere. A country that chooses to do so when its rivals do not may find itself at a serious disadvantage when it comes to retaining talent (though you could make the opposite argument…) or building the sort of technology that allows it to uphold its values, as we discussed last week. Second, and more simply, we don’t know how to improve living standards without a growing economic sphere. As Torsten Bell put it this week:
Whenever people say we need to move beyond caring about GDP, remind them it’s actually the truly dreadful lack of it that has seen our household incomes stagnate for the past decade.
Indeed.
Quick links
  1. Duck, duck… bunny? Computer vision model confused in the same way as humans by optical illusion!
  2. Nation state OKRs. How China plans to measure its performance over the next five years, in balanced scorecard format.
  3. Experts are back! Public opinion has (surprisingly?) swung back in favour of experts over “the wisdom of ordinary people”.
  4. Talk less, smile more? Your conversations are the wrong length, study suggests.
  5. Model ambassador. DeepMind researchers build a model that teaches itself the game Diplomacy. (See also this thread on the surprising amount of academic work there is on Europa Universalis IV)
BONUS: The final part of my Interintellect salon series on “lessons from the Middle Ages” is next Tuesday, on war and violence. Tickets available.
The bit at the end
Thanks for reading all the way through.
Your comments, questions, shares, forwards, etc, always welcome. Just hit reply.
Until next week,
Matt Clifford
PS: Lots of newsletters get stuck in Gmail’s Promotions tab. If you find it in there, please help train the algorithm by dragging it to Primary. It makes a big difference.


Did you enjoy this issue?
If you don't want these updates anymore, please unsubscribe here.
If you were forwarded this newsletter and you like it, you can subscribe here.
Powered by Revue