AI Links, 10/20/2025
Dan Williams on AI and human interdependence; Noah Smith on AI and the economy; Joseph Epstein on battling computers; Amjad Masad on programming physical space
interdependence solves the human alignment problem. It’s how we align human interests.
What happens to such interdependence in a world of superintelligent machines? When there are machine workers that are far more effective and efficient than people—that don’t call in sick, complain about the boss, or start a union? When machines can advance the frontier of human knowledge and innovation without human bias? When they can provide sex and companionship without any of the annoying complexities, conflict, and compromise that characterise human relationships?
This would be a world in which many people, whether wealthy capitalists or ordinary consumers, rely more and more on machines and less and less on other people. It would be a world where the glue holding human solidarity and cooperation together gradually dissolves.
I think of markets as playing a central role in the phenomenon of human interdependence. In this hypothetical AI world, are markets less important, or can we even do away with them entirely?
If AI is really the only thing protecting America from the scourge of Trump’s tariffs, then a bust in the sector could change the country’s entire political economy. A crash and recession would immediately flip the narrative on Trump’s whole presidency, much as the housing crash of 2008 cemented George W. Bush’s legacy as a failure. And because Trump’s second term is looking so transformative, the fate of the AI sector could potentially determine the entire fate of the country.
These days, AI accounts for a lot of new spending in the economy (I would prefer to say that it accounts for a lot of new patterns of specialization and trade). Also, as Noah points out, a lot of financial capital is betting on getting returns from AI, and those returns could be late in arriving or eroded by competition, leading to a financial bust.
The other night while walking I separately ran into three of my former high school students from the early 2000s. They are are well employed, married, with children. All of them use AI at work. One of them said, “I feel like I’m a million times more productive, my co-workers feel a million times more productive, but where is it showing up? I don’t see it in revenue.”
Two weeks ago I attempted to order two shirts from Jos. A. Bank, the men’s clothing store. After roughly 40 minutes on my computer, which sent me back innumerable times to my iPhone to reset my password, username, and everything but my hairstyle, I gave up and called customer service to place my order. A customer service agent took 20 or so minutes to respond to my call, leaving me listening to jumpy music and a robot voice coming on every 40 seconds or so imploring me to stay on the line and reminding me how important my call was to them.
Right now, the only way that I can navigate a web site is to ask ChatGPT for help. Soon, the whole web-design industry will (should) collapse, replacing the web interface with chat interface. And of course the same should go for customer service—it should feel like you are talking to the customer service manager instead of an idiot menu.
And don’t get me started on televisions. We inherited a “smart” TV a few years ago and it has sat idle. Back when televisions were not “smart,” we could figure out how to turn one on and find a channel. No more. It will be a great day when instead of having to figure out all the buttons on the remote(s) we can just have a chat interface.
Interviewed by Balaji Srinivasan, Amjad Masad says,
one thing I would really like to happen, you and I discussed, is like being able to program the physical world. Yes. Because I think we’re we’re headed to to a period where robotics are becoming real. You can like buy a unit dog or unitary robot. Like you should be able to vibe code it.
I enjoyed the whole conversation. As a satisfied user of Replit, I am ready to hear Masad’s optimism.
Pointer from Balaji.
substacks referenced above: @
@
@
@
A few random thoughts:
I spent some time learning a bit about the Econ Nobel Prize winners recently. It strikes me that the “AI is propping up the economy” is exactly the sign of a healthy economy – not everything gradually doing a little better across the board, but a lot of churn in different sectors as creative destruction and innovation play out, with a net effect of ‘everything doing a little better’. It also seems to me that the large hyperscalers have been sitting on a lot of cash for a long time without anything productive to really do with it (e.g., Google’s “Bets”), and the AI boom gives them a more productive place to put that capital. Interesting to watch Apple, which is sitting on a tremendous amount of cash and has not deployed it…
Now, will it be productive? I too feel the paradox of “I feel like it makes me a lot more efficient, but I’m not spending less time at work and were not doing more / hiring less in any dramatic way.” I suspect that we are a little more productive today, marginally, but that “a lot more productive” will require a re-shaping of the way things are done, and we won’t see the maximal gains with just “the way things have been done + AI = more productivity”.
And people like Bernie Sanders complain about AI taking everyone’s job. Will this be a real thing? I tell my kids “please, take all of our jobs! I’d love to have robots doing everything for us!” It won’t happen of course, because it’s never happened. Technological improvements might make our lives better, but they don’t leave us without things to do.
I just purchased a new washer/dryer. The old one was repairable, but I’d either have to pay someone essentially the same price as a new one to do the repair, or spend a day plus multiple orders online / trips to some random store with spare parts to repair it myself. That’s a problem on its own – I’d love things to be easily repairable. I bought the most basic model I could find of a nice brand, rather than what I’d usually done which is to buy the one with the most features. I think we’re at a place where simpler is going to be a *feature* of devices, rather than features being features. So-called smart TVs are the chief example. Simplicity is difficult to pull off…
“And don’t get me started on televisions. We inherited a ‘smart’ TV a few years ago and it has sat idle. Back when televisions were not ‘smart,’ we could figure out how to turn one on and find a channel. No more.“
I love it when Arnold goes all in on his Andy Rooney impression. Instant classic.
The television layout isn’t a hidden IQ test. Find the app you want and then click on it. You’re done. Our 11 yo does just fine in navigating to where she wants to go. And, it’s much easier than programming the VCR clock from back in the 80s.
https://youtu.be/3vh6ojj8YKg?si=n5u18IQ-JE2yZ9Xa