Noah Smith

Noah Smith is the economics writer behind Noahpinion and a former finance professor. His AI commentary examines how automation changes tasks, specialization, and wages. He has argued that even AI capable of outperforming humans at every task could leave room for well-paid human work, while acknowledging uncertainty about employment outcomes.[2][3][5]

On December 1, 2022, Smith and roon published “Generative AI: autocomplete for everything”. They predicted that generative AI was more likely to complement and empower workers than broadly displace them, although some people would lose out. Their argument focused on tasks within jobs: automation could take over parts of a worker’s role while making the remaining work more productive.[4]

Smith and roon called their proposed human–AI collaboration the “sandwich” workflow: a human supplies intent and a prompt, AI generates options, and the human selects, edits, and finishes the result. They used Copilot and Midjourney to illustrate the division between human direction, machine generation, and human correction. Smith imagined a writing workflow in which he would supply bullet points, AI would expand them in his style, and he would revise the prose and add details and links.[4]

In “Plentiful, high-paying jobs in the age of AI”, published March 17, 2024, Smith accepted the possibility that AI could become better than humans at every conceivable task. He nevertheless argued that good-paying human jobs could remain possible, perhaps likely, because comparative advantage concerns relative specialization across one’s own possible tasks rather than absolute superiority over another worker.[5]

His concrete example was a venture capitalist who types faster than a secretary but still hires one because making deals is a more valuable use of his time. Being better at everything does not require doing everything. Smith and roon had used the illustration in their joint essay; Smith later credited its provenance to Greg Mankiw’s economics textbook and, through Mankiw, Paul Samuelson.[5][4]

On July 20, 2025, Smith argued in “Stop pretending you know what AI does to the economy” that AI had not yet had a detectably negative effect on the job market. He explicitly left open the possibility of future mass job losses and sharply rising inequality: “We don’t know.”[6]

Smith guest-hosted the May 18, 2026 Latent Space episode with Brandon Anderson, interviewing The Fourth Law founder Yaroslav Azhnyuk about autonomous drones. Smith pressed for distinctions between guidance and decision-making and asked what counted as mission success and which failure modes could prevent it.[1]

Drawing on his own reported experience with machine learning failing outside its training distribution, Smith asked how computer-vision-guided drones would avoid attacking the wrong thing when fog rolled in. He later distinguished image recognition and tracking from more difficult decision-making with long tails of edge cases. He described terminal guidance as building on capabilities known to work, while stressing that this did not make the engineering trivial.[1]

Before these AI essays and interviews, Smith had moved from academic economics into professional commentary and subscription blogging.[2][7]

Smith earned a physics BS with distinction from Stanford University in 2003 and an economics PhD from the University of Michigan in 2012. He subsequently worked as an assistant professor of finance at Stony Brook University and as an economics columnist for Bloomberg Opinion.[8][2]

Smith introduced Noahpinion on Substack on November 24, 2020, initially alongside his Bloomberg work. He intended it to replace much of his Twitter commentary and provide space for longer posts, technical discussion, and topics outside his opinion columns. He valued blogging’s room for “nuance and elaboration” beyond a 280-character format. In his later account, he says he left Bloomberg in late 2021 to blog full time.[7][2]

Smith describes Noahpinion as a techno-optimist publication: he expects technological innovation to be more transformative than many people recognize and usually, though not always, beneficial. The newsletter approaches economics, technology, geopolitics, and culture through his economics training.[2]