Some interesting things to read the first weekend in June

Dear Friends,

The most interesting essay I’ve read recently is by Chinua Achebe on the concept of  “chi” in Igbo cosmology. The essay, part of a book published in 1975, was shared with me by the brilliant novelist Uzodinma Iweala at an event about ways that AI can make us more human. As Iweala explained in his speech, the idea of chi is that we each have an infinitely powerful spirit-double who stands by us at all times. They listen to us, they guide us, and they can turn against us. In the essay, Achebe writes, “In a general way we may visualize a person’s chi as his other identity in spiritland—his spirit being complementing his terrestrial human being; for nothing can stand alone, there must always be another thing standing beside it.”

The essay is a beautiful explanation of history and culture. It also offers a parable as we enter a world in which some people conceive of AI as something that will stand ever-present beside us. Achebe tells a story of an accomplished wrestler who decides to wrestle in the world of spirits:  

“There he also throws challenger after challenger, including many multiple-headed ones—so great was his prowess. At last there is no one left to fight. But the wrestler refuses to leave. The spirits beg him to go; his companion praise-singer on the flute pleads with him. But it is all in vain. There must be somebody left; surely the famed land of spirits can do better than this, he said. Again everyone begs him to collect his laurels and go but again he refuses. Finally his own chi appears, reluctant, thin as a rope. The wrestler laughs at this miserable-looking contender and moves forward contemptuously to knock him down whereupon the other lifts him clear off the ground with his little finger and smashes him to death.”

My favorite new podcast is a surprising one from the folks at One Little Goat Theatre in which the actor Richard Harte reads Finnegan’s Wake. (The director, Adam Seelig, is also there to explain things from time to time.) It’s a way to experience the book in a way you have never before. 

Moving more toward the present, this is a lovely profile in The New Yorker about Geoffrey Hinton, the “godfather of AI,” a brilliant engineer who designed the architecture of neural networks, and then began to worry about the ways they could upturn the world. I interviewed Hinton earlier today at the UN’s AI for Good Global Summit, which you can view here at 7:36. Are you convinced that AI systems can’t have subjective experiences? He will try to convince you otherwise. (This, by the way, was my favorite of the podcasts I listened to while preparing for the interview.)

On Thursday, jurors delivered a verdict in Trump’s New York trial, and my old boss, David Remnick, was ready with a classic. It’s also tennis season, which means it’s a great time to read this Atlantic essay on the “ruthless, cold-blooded unkillability” of Novak Djokovic. I also enjoyed this interview with Paolo Benanti, the AI advisor to Pope Francis. And one of course needs to read Jennifer Senior on the writing, and pain, of Suleika Jaouad. “It is tempting to look at Suleika’s illness as an origin story, the thing that forced her to live an exceptional life. But another way to think about it is that Suleika is an exceptional person to whom illness happened. Speak with her friends, and you get the sense that she has always lived her life like the rest of us, but in a much larger font.”

This section is sponsored by Elastic, the Search AI Company. Elastic enables everyone to find the answers they need in real-time, using all their data, at scale. Its solutions for search, observability, and security are built on the Elastic Search AI platform—the development platform used by thousands of companies, including more than 50% of the Fortune 500. Visit here to learn more.

When the mathematician Hans Moravec was building robots in the 1980s, he came up with a beguiling paradox that bears his name: “It is comparatively easy to make computers exhibit adult level performance on intelligence tests or playing checkers, and difficult or impossible to give them the skills of a one-year-old when it comes to perception and mobility.” Or, in other words, “What is hard for humans is easy for machines, and what is easy for humans is hard for machines.” It’s easy to teach a robot to play chess, it’s hard to teach a robot to play hopscotch. Robots are great at calculus but they have trouble recognizing faces.

You can guess where this is going: The advancements in neural networks and AI learning are challenging Moravec’s paradox. A team at Carnegie Mellon created a robot dog that moves through the landscape with an eerie precision. The robot has a camera eye and was taught to recognize obstacles through an AI training technique called reinforcement learning—trial and error, basically. Other teams have taught robots by having a human show the robot how to do something—cooking shrimp, for example—and then letting the robot combine the actions it learned to perform extraordinary feats, such as cooking a three-course meal. 

Unlike large language models, which can be trained on the profusion of words found on the internet, this new generation of adaptable robots must be trained on movement, leading researchers to discuss the dream of developing “large behavior models.” (Imagine a kindergarten for robots.) The hope is that AI-trained robots will be able to navigate the chaotic environments that humans often create: the busy floor of a city hospital or the messy living room of a suburban home. (I should add that I also like the idea of non-robotic exoskeletons.) I was particularly taken by a robot named Punyo, which has giant, soft, tactile arms and is being used in what’s called whole body research. It turns out that the height of robotics isn’t some HAL 9000-like omniscience, but rather a gentle robot that’s working very hard to figure out how to open the front door while carrying three big bags of groceries.

Cheers * N

I hope you enjoy this newsletter, which, thanks to your support, now has more than 400,000 readers. Please continue to forward it to anyone else who might enjoy it. They can sign up here.