Resolution will be based off my subjective judgement of the quality of accessible models by each group. Any model I can access will qualify as accessible. Other models (e.g. a Chinese citizen restricted model) will qualify only if I can get a reliable sense of their capabilities.
The models to be considered are likely to primarily focus on chat models, but if e.g. one company develops a personal assistant agent AI which I perceive as closer to AGI than any chat model, I would resolve to that company.
If two listed entities merge I will resolve to both weighted by the relative valuation of each at time of merger. If one is acquired by the other, I will resolve to the entity which did the acquiring.
I will be generous in adding options until EOY 2024, after which I will cease to add options. Feel free to suggest answers.
Related questions
@OnurcanYasar Until a week ago, I would have said OpenAI, hands down. But now that Sam Altman has been fired by OpenAI and then hired by Microsoft, and most of the people working at OpenAI have threatened to leave OpenAI and join him, the obvious leader in the immediate future is going to be Microsoft (it's still possible that OpenAI will salvage things, but the best way for them to do that would be for them to effectively become a subsidiary of Microsoft).
@DanielParker What’s your opinion about meta and google. I struggle to find good criteria to figure out who is ahead. I really hoped for Microsoft stocks getting cheaper but didn’t happen yet, so I can buy into it
@OnurcanYasar Meta and Google are late to the game. They've been scrambling to catch up, but OpenAI has had the lead. All of the major players have enough GPU power to throw at the problem, so the question is who has the best ideas and engineering to improve on the existing designs. In the past that has been the engineers at OpenAI and I don't see any reason for that to change.
@DanielParker Wasn’t there a few months ago a story that google had a LLM developed but didn’t deploy it for some misguided reasoning? But you seem right in general. The only thing that I am not som much sure of is the possibility that LLMs are probably not that much general purpose and that open AI is just crushing one aspect of intelligence as LLMs are neither good chess players nor good in explaining game theory