Séb Krier’s 8 Summer Reads
Your vacation, sorted
While others slip on shades and flop by a suitable body of water to resume smartphone swiping until the dopamine/sunshine cocktail melts them into a burnt stupor, you are here looking for long-reads. In other words, you’re our kind of nerd.
Thankfully, we have a hotline to AI-policy hipster Séb Krier, who agreed to direct our holiday nerding. He even suggests the music: this bumping album from his own YouTube channel. So, apply sunscreen. Fetch an icy beverage. And sun yourself on Séb’s hot takes…
(We’re taking a break ourselves, away till September. But we have plenty of pieces upcoming, including the robotics revolution, AI for history, and an explainer on hallucinations. Stay tuned!)
—Tom Rachman, AI Policy Perspectives
1. Why Facts Don’t Matter When People Shout
Séb Says: The philosopher Dan Williams writes that many of the raging disputes today between political foes are not actually contests over what’s true. Instead, they’re often rooted in what the great 20th-century journalist Walter Lippmann dubbed “pseudo-environments”: people’s self-serving distillations of the world. When I was younger, I thought the reason we get polarization is because we don’t have access to the same facts. This suggests that, with one more bit of debate and better trimming of the media landscape, we’d fix everything. But this piece makes the case that, even with shared facts, we each have an interpretative lens. And reconciling these is much harder. There may also be some benefits to the splintering of realities, because then you get competition in the marketplace of truth. If some group believes that eating laundry-detergent pods is going to cure Covid, you’ll get quick feedback. This doesn’t mean splintering alone is valuable—you still need the right norms and institutions that value accuracy and the free pursuit of knowledge.
How Tribes Construct Rival Realities (Conspicuous Cognition)
2. Explaining Wealth and Poverty
Séb Says: Two talented thinkers—David Oks and Deena Mousa—separately ponder why some developing countries rise while others struggle. David points to human capital: while China tore apart old social systems and built a modern workforce, India failed to update as radically. Meantime, Deena interrogates the persistent difficulties of Malawi, which she ascribes to populist and expensive policies, like fertilizer subsidies, that have blocked economic reform. I appreciate articles that look beyond AI to unpack the institutional, geographic, economic, and societal factors that shape how technology can change society (or not). A lot of people in the Bay Area overrate intelligence as the bottleneck to improvements in society. The world would be simpler if this were true! AI is going to be transformative, but this truism doesn’t tell you how, where, why. Articles like these provide the texture that simplistic models of the world skip.
Why China Got Rich and India Didn’t (David Oks) &
We Don’t Know Why Malawi Is Poor (Under Development)
3. What AI Does to Culture
Séb Says: Here are another two great articles that I’d suggest reading in succession, this time on how AI could influence culture. There’s a Q&A by Anika Meier with the British artist Mat Dryhurst; and a New Yorker column by Kyle Chayka on how Claude’s design “taste” is turning into a cliché. Excellent stuff is happening in the interaction of AI with arts and culture, but it’s often missed. Cultural elites often cite banal talking points bemoaning technological influences as uniformly negative, while tech companies with the aesthetic sensibilities of a jellyfish proudly sponsor the safest, most out-of-date, generic, artists and thinkers. So it’s refreshing seeing someone like Mat Dryhurst, who deeply understands both the tech and art, explore unconventional ideas at length. I’ve also been banging on about things like model multiplicity, customization, and decentralization for a while—all of which would help address the bland commodified pre-packaged designs that AI model providers may otherwise converge on, and which the New Yorker piece eviscerates. Interestingly, some AI labs now appear to be pushing back against design clichés.
Who Shapes Culture? (Sleek) &
The AI Design Aesthetic That’s Taking Over the Internet (The New Yorker)
4. Lessons for Public Health
Séb Says: A further two-part recommendation. First, another deeply researched piece from Saloni Dattani, who looks into how the world produced Covid vaccines in haste, which she attributes to more adaptive regulations, massive public financing, and accelerated trials (among other factors). Second, Brian Potter wrote an unexpectedly gripping history of how a flesh-eating parasite seemed to be under control—but is now back. (Don’t entirely panic: It mainly affects livestock.) While Saloni’s piece shows what we can learn from Operation Warp Speed, the U.S. program to hurry up vaccine development, Potter’s piece highlights the fragility of progress. A lot of people in AI have been exploring agendas like d/acc and societal resilience—that is, advocating for safe technology not by slowing down AI, but by using it to speed up safeguards. This is great, and I hope we can remain laser-focused on (a) the importance of institutional design; and (b) the many environmental, political, and logistical failures that can slow things down. Doing this well means nurturing scientific and economic expertise outside of generalist AI talent too.
Why Were Covid Vaccine Trials So Fast? (The Clinical Trials Abundance Blog) &
The Fall and Rise of Screwworm (Construction Physics)
5. What Yesterday Teaches About Tomorrow’s Tech Transformation
Séb Says: Talk of technology’s last societal overhaul, the Industrial Revolution, is in the air. This piece from Kara Dimitruk & Ben Southwood looks even further back, considering how the overhaul of 17th-century English property rights set up the explosive industrial development that followed. The changed property rights increased agricultural output, allowing more laborers to take urban industrial jobs; opened up land for development into mines and factories and housing; and cut transport costs. Might there be lessons for the AI revolution? Today, the question of how to “reorganize” rights is underexplored. For example, property rights can be too rigid, and heavily forestall growth and prosperity. Yet changing them can seem nightmarish and intractable. But this piece shows how that has been achieved before, what it could unlock, and how political change can sometimes involve win-win solutions that genuinely make all parties better off, rather than being mere slogans. It makes me wonder: what is today’s equivalent of “enclosure with compensation” that consolidated fragmented land holdings or “turnpike trusts” that helped pay for transport infrastructure?
How Smashing the NIMBYs Created Modern Capitalism (Works in Progress)
 | A guest post by
|