Join 📚 Josh Beckman's Highlights
A batch of the best highlights from what Josh's read, .
The **lab mice problem** **Today's "subhuman" AIs are safe.**What about **future AIs with more human-like abilities?**
Today's AI systems aren't advanced enough to exhibit the basic behaviors we want to study, such as deceiving and manipulating humans.
Like trying to study medicine in humans by experimenting only on lab mice.
How We Could Stumble Into AI Catastrophe
Cold Takes
My gut tells me this pattern, which doesn’t really have a name, is going to be *the* pattern for working with these technologies to do complex things. The best name I can find for it is “specification repair”—see, for example, [this paper](https://arxiv.org/html/2309.16120v3). You ask a question and the LLM transforms it into a more detailed specification of a problem, then writes code to test that specification, then creates more specification, then tests that.
Is “Specification Repair” the AI Endgame?
Paul Ford
Interlace is “the device of interweaving of a number of different themes … all distinct and yet inseparable.” The device is thought to have originated with Ovid. Denis Feeny in his introduction to Ovid’s Metamorphoses, notes that the “haphazard chain of association is entertaining, but it also reinforces the Ovidian theme of the very contingency of connectedness.”
Crossing the Sunshine Skyway
Robin Sloan
...catch up on these, and many more highlights
Readwise