A.I. Is Different From What You Think with Michael Fraser | HN 241
Adam welcomes to Haman Nature Michael Fraser, co-founder of Action/Insight and a man very familiar with both Austrian economics and the inner workings and capabilities of the Large Language Models (LLMs) we are calling Artificial Intelligence these days. Michael has a fascinating claim for us to explore. Are these LLMs more than "just" machines? Are they Misesian actors? Let's find out!00:00 — Intro. Welcoming a smart and fascinating fellow, Michael Fraser! Who is this guy and why are we listening to him about LLMs and Misesian action?09:00 — Michael lays out his thesis: LLMs (regardless of any theory of consciousness) are Misesian actors in the world. Mises's theory has broader applications than are required to understand markets.14:40 — What things are Misesian actors and what aren't? Is there a spectrum? Are there levels? Calculators? Beavers? Dogs (mine or yours?), LLMs? Politicians? What is Misesian action? How do we evaluate teleology with regard to LLMs?27:40 — Time, interaction, iteration, trajectory: how do these concepts relate to this issue? How should we think of LLMs? How many "Claudes" or "Groks" are there, could be, or have ever been?35:33 — How does math (and reality itself?) constrain how LLMs exist and act in the world? How does path dependency relate to this question?41:20 — How do we know Michael isn't just being fooled by a fancy word model and hallucinating Misesian action? How do Mises's "alter ego warrants" relate to us being fooled about "personhood"?47:19 — On a Misesian action spectrum, between calculator and humans, where should we place beavers, dogs, chess engines, LLMs, Teslas, and your cousin Ernie? Are there levels? Does qualia have anything to do with this?53:02 — How should we understand LLM "hallucinations" within Michael's framework? Are LLMs "answer machines" or "useful cauldrons"?59:02 — How should we think about the Turing Test vis-a-vis LLMs?1:04:44 — How should we think about the terms AI, AGI, ASI, etc? Are these terms meaningful?1:09:30 — Why does interaction with another matter? Surely Misesian action occurs when we act alone, right? What does the classic Robinson Crusoe economics case reveal here?1:13:09 — Do we need an "LLM rights tribunal" because of the way we treat LLMs?1:18:27 — Does Adam buy Michael's argument?1:19:58 — What does Michael think is happening inside an LLM?1:30:30 — Outro. Thanks for watching Haman Nature!Michael on X: https://x.com@zeroshotnothing [https://x.com@zeroshotnothing]Michael's paper presented at the AERC: https://austrianintelligence.org [https://austrianintelligence.org]Michael's business: https://actioninsight.ai [https://actioninsight.ai]Michael's interview with Bob Murphy on the Infineo podcast, explaining his recent paper "Artificial Action" that he presented at the Austrian Economics Research Conference:
Michael's previous appearance on the Infineo podcast:
Michael on the Bob Murphy Show:
Sponsor mentioned: https://TheSwanBrothers.com/crossover [https://TheSwanBrothers.com/crossover]Haman Nature Merch Store: https://shop.hamannature.com [https://shop.hamannature.com]Adam on X: https://x.com/Rerazer [https://x.com/Rerazer] Bob on X: https://x.com/BobMurphyEcon [https://x.com/BobMurphyEcon] The Bob Murphy Show: https://www.bobmurphyshow.com/ [https://www.bobmurphyshow.com/] Adam’s other show: https://naturalorderpodcast.com [https://naturalorderpodcast.com]
This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit hamannature.substack.com/subscribe [https://hamannature.substack.com/subscribe?utm_medium=podcast&utm_campaign=CTA_2]
