Very good post and I agree with the trend you’ve identified (and the main thesis that we will not have access to the frontier, now, or ever again, under current trajectory), but I would qualify this a little better:
”Since the release of ChatGPT, at any given time, anyone on the planet with a few bucks could access the current most capable AI model, the SOTA.[4]” because in its current state, I’d say that’s mostly inaccurate. We know that labs have had better models than what they’ve released as the SOTA, basically every generation. There’s no reason or real way that could not be the case, but this (restricting based on safety, and the gap being so large), is likely the bigger change.
Very good post and I agree with the trend you’ve identified (and the main thesis that we will not have access to the frontier, now, or ever again, under current trajectory), but I would qualify this a little better:
”Since the release of ChatGPT, at any given time, anyone on the planet with a few bucks could access the current most capable AI model, the SOTA.[4]” because in its current state, I’d say that’s mostly inaccurate. We know that labs have had better models than what they’ve released as the SOTA, basically every generation. There’s no reason or real way that could not be the case, but this (restricting based on safety, and the gap being so large), is likely the bigger change.