Cal Newport
speaker
22,298 appearances
39 recordings
5 series
first heard Oct 2025
last heard 10 Sep
Cal Newport’s voice in public audio — every appearance, attributed to the second.
Trend
recordings per month · last 12 monthsRecordings per month over the last 12 months — 39 in all, peaking in Jun 2026 with 9.
Appearances
Because it turns out, one of the only ways, if you're going to let an LLM come up with all the plans for a prompt loop or an agent, which I think is a really bad idea, but if you're going to do that, one of the core ways you can figure out what's going on is if it's reasoning about each move in English and tokens in English, like here's what I'm going to do next, well, you could have language style tools
podcast-host
that look at these traces and are like, ooh, this looks dangerous.
podcast-host
Why don't we stop the prompt loop?
podcast-host
This is a widely accepted idea right now.
podcast-host
I'm going to bring a paper up here on the screen.
podcast-host
The title is Chain of Thought, Monitorability, a New and Fragile Opportunity for AI Safety.
podcast-host
It has like
podcast-host
All the names on his co-authors from the companies, from the safety community.
podcast-host
It has like Jeff Hinton and like Ilya Sutskever, as they call them, expert endorsers, people who are reading it.
podcast-host
Let me just read you the abstract of this paper.
podcast-host
This is a recent paper.
podcast-host
AI systems that think in human language offer a unique opportunity for AI safety.
podcast-host
We can monitor their chains of thought.
podcast-host
for the intent to misbehave.
podcast-host
Like all other known AI oversight methods, chain of thought monitoring is imperfect and allows some misbehavior to go unnoticed.
podcast-host
Nevertheless, it shows promise and we recommend further research in the COT monitorability and investment in chain of thought monitoring alongside existing safety methods, right?
podcast-host
So there's this idea that makes a lot of sense in the AI security community.
podcast-host
If you're going to power agents without a LIMS,
podcast-host
We could look at the chain of thought, which helps them get better results, but they also has the side effect of it gives us some insight into what they're doing.
podcast-host
And if we see things in there like kill all the humans, you know, we're like, oh, maybe we should stop what this agent is doing.
podcast-host
Showing 301–320 of 22,298 · page 16 of 1115
← Previous
Next →