Pluribus didn't solve poker. It's limited to fixed starting stack sizes. It can't exploit weak opponents, it tries to approach a Nash equilibrium, but in multiplayer poker, Nash equilibrium doesn't have the theoretical guarantees it does in head's up. And lastly, it requires a ton of compute.
Curious about how much risk Meta leadership was comfortable with when they decided to layer Yann. Perhaps the winds of open research were already blowing a different direction at the company, and he had already indicated that he wanted to leave as a result of that. We can only guess.
Kind of hilarious to me to consider him "failing" with LLMs. Given his remit was a research time horizon of 8-10 years, and the fact that he's gone on record saying that he expects the technology will stall out in the time horizon, it seems he can only take Ws and ties. Indirect influence on open-sourcing the models to propel research forward (which is pretty important for a chief scientist) which added benefit for Meta's other products.
Also used to be in that world and identify similarly in terms of my lack of love for gambling.
I'd suggest that you're empirically incorrect in saying that there is no perfect approach against a skilled player (6handed games which often reduce to a single heads-up interactions by showdown):
1. we know that a Nash equilibrium exists for every two-player zero-sum game such that it’s mathematically unexploitable
2. Pluribus approximated the Nash well enough (didn’t have to search over 10^161 possibilities) to crush high stakes skilled player over a good run of hands
You're not wrong that knowing the odds is a component of the skill, but to suggest that skill in poker stops there is minimizing many of the advanced aspects that require playing at a higher level (information management, assessing a player's likely range, determining the equity of a player's range with cards to come, realizing when your or their range is capped, etc)
I feel one of the most useful skills picked up by poker that people don't explicitly speak about is managing your information effectively.
Deceiving my opponent has the connotation of this happening in one instance. After you realize that you can't convincingly deceive your opponents in poker into perpetuity, it becomes a game of managing your image —revealing the right information while being conscious of information that you shared in the past (if you're playing someone skilled or perceptive, that is).
On the flip side, what an excellent game to help people pay attention to signals, figure out how to weigh them appropriately, and appropriately act on them when the situation calls for it.
As someone focused on grad school, I find myself much less frequently getting stumped by problems and rage-quitting. I also use some prompts which help to speed up my learning in general while making sure any LLM doesn’t give me the answer.
One of my research interests is on how humans use expert systems (akin to how Go players’ ELO ramped significantly after the release of AlphaGo).
reply