The Academics That Think ChatGPT is BS (Better Offline Podcast) - I didn’t take detailed notes on this one, and probably should have, because it was packed with very insightful and well thought out hypothesis’ about how LLMs work but also about how we should view them, whether they are actually thinking, how they differ to human learning, and how they are impacting academia. These researchers highlight some of the most important things to consider, such as lying vs bullshitying vs hallucinating, and the danger of anthropomorphising them. It vears into philosophy and psychology, but grounded in computer science and mathematics. You'd think these boffins would have the whole thing figured out, but even these guys occasionally run into difficulty describing what’s going on. LLMs seem to be teaching us humans about our own cognition and consciousness. It’s the sort of episode that should be required listening when using these tools, they are powerful but could be detrimental to your abilities if used in the wrong way long term. IMO, use them as a booster rather than a crutch, imagine they could be taken away at anytime, you still need to be able to function at a high level without them. podcastindex.org #