Couldn't agree more with your points. Another piece to this is that yes generative AI can generate a ton of output (so it's highly efficient), but that doesn't mean that any of it is actually valuable (so it's ineffective).
I've been working on an applied project synthesizing the ideas of the Judgment Based Approach specifically for busy executives. I've combined it with my retrospective experience working "in the trenches" at Uber and Lyft where I often found myself to be "the one economist in rooms filled with data scientists." Here's an article from that series that overlaps very well with yours here. https://www.economicsfor.com/p/entrepreneurial-judgment-and-ai-precision
(P.S. You may remember me from Mises U and AERC years ago. Last we saw each other was in Reno at the Hayek Group talk you gave hosted by Mark Pingle where Mark Packard also attended. My younger brother Steven Spurlock was a student at this year's Mises U too).
Peter, may I take you on a slight tangent? C. S. Peirce defined three types of reasoning: deductive, inductive, and abductive. Deductive reasoning is the essence of LLM operation over massive stored data. Both inductive and abductive are ampliative reasoning, when the boundaries of the prior data are breached. I wonder if AI can eventually (soon?) do inductive reasoning. I believe that the initial conjectures, hypotheses, or hunches that populate abductive reasoning must remain vested in human agents, like entrepreneurs.
Great points. I’m actually working on a paper on induction (as a method used by entrepreneurs in “judging” the future). Not sure if agents can do induction but I agree with you that abductive reasoning seems outside their capabilities.
Couldn't agree more with your points. Another piece to this is that yes generative AI can generate a ton of output (so it's highly efficient), but that doesn't mean that any of it is actually valuable (so it's ineffective).
I've been working on an applied project synthesizing the ideas of the Judgment Based Approach specifically for busy executives. I've combined it with my retrospective experience working "in the trenches" at Uber and Lyft where I often found myself to be "the one economist in rooms filled with data scientists." Here's an article from that series that overlaps very well with yours here. https://www.economicsfor.com/p/entrepreneurial-judgment-and-ai-precision
(P.S. You may remember me from Mises U and AERC years ago. Last we saw each other was in Reno at the Hayek Group talk you gave hosted by Mark Pingle where Mark Packard also attended. My younger brother Steven Spurlock was a student at this year's Mises U too).
Cameron, great to hear from you and thanks for telling me about your project.
Peter, may I take you on a slight tangent? C. S. Peirce defined three types of reasoning: deductive, inductive, and abductive. Deductive reasoning is the essence of LLM operation over massive stored data. Both inductive and abductive are ampliative reasoning, when the boundaries of the prior data are breached. I wonder if AI can eventually (soon?) do inductive reasoning. I believe that the initial conjectures, hypotheses, or hunches that populate abductive reasoning must remain vested in human agents, like entrepreneurs.
Great points. I’m actually working on a paper on induction (as a method used by entrepreneurs in “judging” the future). Not sure if agents can do induction but I agree with you that abductive reasoning seems outside their capabilities.