Steve Newman on coding with agents; Tim B. Lee on the Anthropic-Pentagon dustup; Noah Smith says that AGI is here now; Dan Williams on AI vs. Social Media
This analysis can easily (and sadly) be extended to incorporate AI vs. the legacy media.
For example, during the recent ICE shootings in Minneapolis, I had far more productive, dispassionate and enlightening conversations with Gemini than anything I viewed from the coverage and conversations on either CNN or Fox News. It is very difficult to do thoughtful analysis without ideological bias entering the equation.
With AI, it feels more like the rider is in control as opposed to the elephant.
On the Dan Williams article/quote. I agree completely and suspect this will have a dramatic effect on how Substack and opinion pieces work. It is getting real easy to cut through the BS. As an example, we have argued about the causes of the gender wage gap for decades. Ask AI, and it gives an economically literate explanation and cites the supporting literature.
This has some real potential to elevate the level of discussion and debate.
"I bet that if AI had A) permanent autonomy and long-term memory, B) highly capable robots, and C) end-to-end automation of the AI production chain, it could defeat humans and take control of Earth today."
Except we don't have most of these things, maybe none of them. And unlike AI advances, some of these are more like cold fusion - less than ten years away for the last 20 years.
It is both true that DOW should insist on full and independent control over the details of how anything it procures is used, and that it was very wrong and very dumb to have treated Anthropic the way they did after signing a contract including those restrictive terms.
Just from a libertarian perspective, wouldn’t the stance be that anthropic can choose whether to sell its product to a buyer incumbent on certain restrictions being agreed to?
Would your thoughts change if selling to England? To Canada? To Iran (or does the answer depend on where company is based)?
And how exactly could Claude change the institutions ?
Perhaps "sell" is the wrong word- if, instead, it is "license" you might be on firmer ground. To give an example, "I will sell you this fork but you can't use it to eat meat of any kind," would not find much true libertarian support.
Makes sense. The other analogy I’m thinking of albeit imperfect is selling the cake to a marriage you don’t support. But in that case the issue if I recall was more about being required to sell a specifically custom requested cake design rather then the bakery restricting the use of its generic cake.
You might be on firm ground if it were a matter of copyright or patent and the customer was using it to make a product for sale. That doesn't seem applicable and I don't see another path to restricted use.
Here's a CA house built by ai + bots + 3d concrete printer, list price $375,000 in Yuba, CA (North of Sacramento; 6-7 times annual median $60k income). Took 24 days, more are coming here. And elsewhere.
Functioning markets depend on parties abiding by contracts. If it is so obvious that a company shouldn't tell the military what to do, then why did it sign a contract that supposedly does that? I don't understand why people put the blame on Anthropic here. Let the gov abide by its contracts or go to the market for different ones.
You have taught me how important AI instruction is. I find it much more difficult but rewarding to write an article using heavy AI instruction. It might take me an hour
more, but I learn more about myself in the writing process. I think this applies to teaching in high school let’s say.it makes the students think more.
Re: AI vs. Social Media
This analysis can easily (and sadly) be extended to incorporate AI vs. the legacy media.
For example, during the recent ICE shootings in Minneapolis, I had far more productive, dispassionate and enlightening conversations with Gemini than anything I viewed from the coverage and conversations on either CNN or Fox News. It is very difficult to do thoughtful analysis without ideological bias entering the equation.
With AI, it feels more like the rider is in control as opposed to the elephant.
"LLM’s responses are generally much more accurate, evidence-based, and in line with expert consensus than what you get from most social media posts."
Social media posts??? Isn't that obvious? I'd think it would even be better than most mainstream media news items.
On the Dan Williams article/quote. I agree completely and suspect this will have a dramatic effect on how Substack and opinion pieces work. It is getting real easy to cut through the BS. As an example, we have argued about the causes of the gender wage gap for decades. Ask AI, and it gives an economically literate explanation and cites the supporting literature.
This has some real potential to elevate the level of discussion and debate.
Potential? I suppose.
Likely? No.
The reasons men make more money are many and messy. It is much easier to blame it on discrimination. Easy wins.
"I bet that if AI had A) permanent autonomy and long-term memory, B) highly capable robots, and C) end-to-end automation of the AI production chain, it could defeat humans and take control of Earth today."
Except we don't have most of these things, maybe none of them. And unlike AI advances, some of these are more like cold fusion - less than ten years away for the last 20 years.
It is both true that DOW should insist on full and independent control over the details of how anything it procures is used, and that it was very wrong and very dumb to have treated Anthropic the way they did after signing a contract including those restrictive terms.
Just from a libertarian perspective, wouldn’t the stance be that anthropic can choose whether to sell its product to a buyer incumbent on certain restrictions being agreed to?
Would your thoughts change if selling to England? To Canada? To Iran (or does the answer depend on where company is based)?
And how exactly could Claude change the institutions ?
Perhaps "sell" is the wrong word- if, instead, it is "license" you might be on firmer ground. To give an example, "I will sell you this fork but you can't use it to eat meat of any kind," would not find much true libertarian support.
Makes sense. The other analogy I’m thinking of albeit imperfect is selling the cake to a marriage you don’t support. But in that case the issue if I recall was more about being required to sell a specifically custom requested cake design rather then the bakery restricting the use of its generic cake.
You might be on firm ground if it were a matter of copyright or patent and the customer was using it to make a product for sale. That doesn't seem applicable and I don't see another path to restricted use.
Here's a CA house built by ai + bots + 3d concrete printer, list price $375,000 in Yuba, CA (North of Sacramento; 6-7 times annual median $60k income). Took 24 days, more are coming here. And elsewhere.
https://nypost.com/2026/03/08/us-news/california-3d-printed-home-completed-in-just-24-days-in-yuba-county/
Chris Blattman is an excellent development economist, author of the fine book: Why We Fight
He recently changed from skeptic to aigent vibe coding exemplar, noted by both Steve N & Noah S.
Looks like what an Arnold Kling of TCS might have done, and very useful to researchers.
Here’s his Claude Blattman site https://claudeblattman.com/
2026 will be the year of aigents that really do things. Might be pushing the big foundation models to find more monetization methods.
Certainly makes me feel like spending more time vibe coding now, despite no job to make money, rather than so much blog post reading & commenting.
Functioning markets depend on parties abiding by contracts. If it is so obvious that a company shouldn't tell the military what to do, then why did it sign a contract that supposedly does that? I don't understand why people put the blame on Anthropic here. Let the gov abide by its contracts or go to the market for different ones.
You have taught me how important AI instruction is. I find it much more difficult but rewarding to write an article using heavy AI instruction. It might take me an hour
more, but I learn more about myself in the writing process. I think this applies to teaching in high school let’s say.it makes the students think more.