I love the idea that petertodd and Leilan are somehow interrelated with the archetypes of the trickster and the mother goddess inside GPT’s internals. I would love to see some work done in discovering other such prototypes, and weird seemingly-random tokens that correlate with them. Thigs like the Sun God, a great evil snake, a prophet seem to pop up in religions all over the place, so why not inside GPT as well?
Valentin Baltadzhiev
Karma: 43
A list of all the deadlines in Biden’s Executive Order on AI
[Question] What experiment settles the Gary Marcus vs Geoffrey Hinton debate?
[Question] Can someone explain to me what went wrong with ChatGPT?
[Question] Why do we need an understanding of the real world to predict the next tokens in a body of text?
[Question] Can any LLM be represented as an Equation?
Glad to hear that!
For what it’s worth it’s probably a good thing that the Bing chatbot is like that. The overall attitude towards AI for the last few months has been one of unbridled optimism and people seeing a horribly aligned model in action might be a wake up call for some, showing that the people deploying those models are unable to control them.
You make interesting points. What about the other examples of the ToM task (the agent writing the false label themselves, or having been told by a trusted friend what is actually in the bag)?
On the bright side Connor Leahy from Conjecture is going to be at the summit so there will be at least one strong voice for existential risk present there