alt Hacker News> taking careful precautions against feeding the findings back into training so as to not risk shaping Argon’s reasoning to evade our monitoring. We strongly encourage the rest of the industry to preserve reasoning transparency in these pivotal moments of increased capabilities while navigating alignment risks, so that model thoughts remain helpful in identifying and diagnosing misalignment.
This is good, but they're the slow mover due to this exact thing.
Google is getting punished for not letting the models enter an echo chamber and go faster than humanly possible.
Replies
I don't mind companies taking a more careful approach (= being slower); for one, Google is funding itself, Anthropic and OpenAI rely on a steady influx of investor money while they're not profitable - and I think there's a high chance one or both of them will go under and be subsumed into another player. That player will (I think) most likely be a Google or Microsoft, as they have their own means to buy up a party like that.
Or they may choose not to because the companies are massively overvalued and they have their own AI technologies / technologists already.
google does not return real chain of thought via the API. you can't monitor it.
they use a small model to make fake chain of thought and return that.
google has access to the real chain of thought.
Mmh ok. How much theoretical speed or 'intelligence' gain is realized by allowing reasoning to occur in some inscrutable intermediate representation? Has this been actually tested, how much is it slowing them down, and compared to whom exactly?
OpenAI is the company that originally proposed and popularized chain-of-thought monitoring: https://openai.com/index/chain-of-thought-monitoring/
So no, Google is not being punished, nor are they the people behind this technique.