Amie Pollack
@amiepollack
Amie Pollack
- BabyAI: A simple, two-dimensional grid-world in which the agent has to solve tasks of varying complexity described in natural language.
- Crafter: A Minecraft-inspired grid environment where the player has to explore, gather resources and craft items to ensure their survival.
- TextWorld: An entirely text-based game with no visual component, where the
Jack Clark • Import AI 393: 10B distributed training run; China VS the chip embargo; and moral hazards of AI development
Speaking of past OTs, a number of you have inquired about “the archive” or “reading past issues” and I am here to deliver the news that the inbox is your archive and there is no way to read past issues except to reply to this email and request a past issue. There are too many of you (and too many issues) for me to forward multiple issues or every p... See more
OT 98: IS SHOPPING REALLY IMPORTANT FOR STYLE?!
I offer up this recipe for one of my favorite holiday treats. It is utterly forgiving - you might want to add a bit of black pepper or perhaps some fresh rosemary or thyme. Like spice? Try some espalette pepper powder, a shower of fiercer pepper flakes, or a dollop of mustard.
Whatever you choose, these little crackers, hot out of the oven, go perfe... See more
Whatever you choose, these little crackers, hot out of the oven, go perfe... See more
[12112024, Latent Space @ NeurIPS] Reasoning
docs.google.comLanguage models have randomness built in, the reasoning process should mirror that. humans do not.
Some group of people must exist who are educated, aware, motivated, and at the bleeding edge of a field…who nonetheless have some spare hours in the workweek to spend innovating on something instead of slavishly executing a task.
It’s necessary to be slightly underemployed if you are to do something significant.
In other words, if AI replaces all our jobs, it might not because because what we do is easy, or because AI is so good that it can do as well as we do. It may be because it never mattered if we were particularly good at it in the first place.
We think that we surround ourselves with our interests, but causality often runs the other way: We become interested in what we’re surrounded by.
1) It must be exhausting to read things in order to get mad at them. 2) The social-media internet was built around finding things to get mad at. 3) That motivation ultimately made everything bad. 4) I don’t want to do that; I want people to read newsletters (or articles) because they want to be surprised and delighted by them, because we’re on simi... See more
