ContraMinds Podcast - Unlocking Personal Growth and Professional Excellence · Swami, ContraMinds Labs

Thinking With Machines: Sense-Making, AI Agents, Future of Work - Prof. Vasant Dhar (#063)

·49 min·3 clips

Dhar maps trust in AI using predictability and the cost of error.

1. ContraMinds Podcast #063 focuses on Vasant Dhar’s book Thinking with Machines and AI’s impact on sense-making, work, and trust. 2. Swami interviews Vasant Dhar, a professor at the Stern School of Business, an AI researcher, and the host of Brave New World. 3. The episode asks what humans should do when machines can talk, reason, and produce usable answers without clear visibility into their inner workings. 4. Dhar says modern AI is a black box because it was built to be conversational in our own image. 5. He argues that a major shift came after the 2017 transformer paper and the release of ChatGPT, when anyone could talk to a machine about anything at any depth. 6. He says the old distinction between expertise and common sense broke down, which made AI accessible to everyone. 7. He explains that next-word prediction became powerful because it forced models to learn a great deal about the real world. 8. Dhar says the machine is seductive because it usually sounds right, which makes it easy to pass along outputs without checking them. 9. He keeps returning to sense-making as the discipline of asking whether an output is true, meaningful, and appropriate. 10. The healthcare example shows his point about the worst of both humans and machines, with patients moving through an assembly-line process and little useful data being captured. 11. He says scorekeeping matters in teaching, medicine, and trading because computers can track activity better than humans can. 12. He describes the Damodaran bot as a system that uses many agents, including data acquisition, comparable-company analysis, report writing, and a critic. 13. Dhar says the project failed when they first tried to feed Damodaran’s writings into a model and ask it to learn valuation directly. 14. He says the breakthrough came from breaking valuation into parts and asking each agent to do one task in English. 15. He says the system can take a stock symbol, ask whether the user has a thesis, and produce a five- to ten-page report in about 15 minutes. 16. Dhar says scenario analysis is where such tools become useful because users can change assumptions and see how a valuation shifts. 17. He presents the trust heat map as a way to place problems by predictability and the cost of error. 18. He says AI will create a bifurcation of humanity between people who become superhuman through knowledge and people who become dependent on the machine. 19. The conversation is an interview format with long, explanatory answers from Dhar and frequent examples from finance, healthcare, and education. 20. People interested in AI strategy, valuation, and management will get the most from it, while listeners wanting a light or purely technical episode may skip it.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from ContraMinds Podcast - Unlocking Personal Growth and Professional Excellence