Orlando Stoics

Forget what is inside. The question is not whether a machine feels anything but how fast it could arrive, and whether anyone outside the building would notice. Two of our three thinkers debated this in public. We begin this week with Eliezer Yudkowsky. In 2008 he and Robin Hanson argued this out on Overcoming Bias, and again on stage at Jane Street in 2011. His claim is that progress here is not smooth. A system that can improve its own design crosses a line rather than climbing a hill. The returns feed back on themselves, so the curve bends while nothing outside has changed to warn anyone. The idea goes back to I. J. Good in 1965. Yudkowsky added the part that matters for planning: the first group across has no reason to tell anyone, and a lead of weeks is enough. His 2025 book with Nate Soares, If Anyone Builds It, Everyone Dies, states the case plainly.

We then turn to Robin Hanson, who rejects the premise. Intelligence is not one lever. Capability is a large stock of particular knowledge spread across an economy, and no single system absorbs it by getting cleverer about itself. He also has an outside view: no team has ever pulled far enough ahead on a major technology to outrun everyone else. In The Age of Em (2016) the transition is quick but still economy-wide. In "I Still Don't Get Foom" (2014) he says he cannot find the argument he is meant to be answering. If he is right, capability shows up as spread, not as one lab vanishing over the horizon.

Finally we will discuss Thomas Wolf, who is chief science officer at Hugging Face and answers the question from inside the machinery. In "The Einstein AI Model" (March 2025) he replies to Dario Amodei's claim that we will soon have a country of geniuses in a data center. Wolf's counter is that we will get a country of yes-men on servers. Models are trained to predict the likely answer, and the hard part of science is not answering but asking. He argues from his own experience: he was a straight-A student who turned out to be an ordinary researcher, because school rewards conformity and discovery requires breaking with it. This cuts at the step every fast story depends on, AI doing the research that builds the next AI. He also runs the hub where the weights are posted and counted, which is what lets anyone check whether capability is concentrating or spreading.

Links
Eliezer Yudkowsky

This is our IN-PERSON Stoic discussion and reading (no Zoom link available). We meet every 2 weeks on Saturdays.

READING

# Reasons Not to Worry: How to Be Stoic in Chaotic Times

In this thoughtful and approachable book, journalist Brigid Delaney spends a year exploring how Stoic philosophy can be applied to modern life. Blending personal experience with humor and reflection, she turns to ancient thinkers like Seneca, Epictetus, and Marcus Aurelius to help answer some of life’s most persistent questions—how to find peace, handle uncertainty, and focus on what truly matters.
Rather than presenting Stoicism as abstract theory, Delaney tests its ideas in real time. She compares her own tendencies—impatience, anxiety, and a fast-paced modern lifestyle—with the Stoic emphasis on reason, acceptance, and control. Through this process, she examines how ancient wisdom can be used to navigate everyday challenges, from stress and insecurity to grief and decision-making.
The result is a practical and often humorous exploration of how to live more deliberately. By learning to let go of what cannot be controlled, reflect on what is important, and approach life with greater awareness, Delaney shows how Stoicism can help restore a sense of balance and perspective—even in uncertain or chaotic times.

Book Club Format