Evidence receipt / uncertainty
Published · transcript-backedTrenton Bricken: uncertainty
28 Mar 2024 Dwarkesh Podcast Sholto Douglas & Trenton Bricken — How LLMs actually think
“Totally. It's just that some people will hail chain-of-thought reasoning as a great way to solve AI safety, but actually we don't know whether we can trust it.”
Source trail
Everything needed to verify it.
- Speaker
- Trenton Bricken
- Attribution
- Verified speaker
- Claim type
- uncertainty
- Recorded
- 28 Mar 2024
- Publisher
- Dwarkesh Podcast
Transcript context
…But isn't this how humans think as well? There are the famous split-brain experiments where for a person who is suffering from seizures, they cut the thing that connects the two halves of the brain. The speech half is on the left side so it's not connected to the part that decides to do a movement. So if the other side decides to do something, the speech part will just make something up and the person will think that's legit the reason they did it. Totally. It's just that some people will hail chain-of-thought reasoning as a great way to solve AI safety, but actually we don't know whether we can trust it. How does that change with AI agents, this landscape of models communicating to themselves in ways we don't understand? Because then it's not just the model itself with its previous caches, but other instances of the model.…
Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.