At the very least, we see there is minimal practical value for LLMs for any serious work. This is sort of good news. The effort to build this type of "AI" is al

At the very least, we see there is minimal practical value for LLMs for any serious work. This is sort of good news. The effort to build this type of "AI" is all in the training data and navigating politics. That leaves two possibilities: either another AI winter comes as people fail to capture long term value, or we get less swampy models that are much more useful and trained the correct way.