There is something called tacit knowledge that cannot be easily learned through LLMs. LLMs are currently very good for research and theoretical knowledge. For e

There is something called tacit knowledge that cannot be easily learned through LLMs. LLMs are currently very good for research and theoretical knowledge. For example, learning how to make mechanical watches involves a lot of hands on, tacit knowledge. That skill would also be transferable to building other precision instruments. I don’t think we can currently learn these kinds of skills easily with the help of LLMs. Even if you know everything about bicycles their mechanics, components, and how they work you still have to learn how to actually build, repair, or ride one through practice. Knowing, understanding, learning, and practicing are completely different things. There are also niche areas of expertise that can take years to develop,not just to the point where you know the terminology and jargon, but where you understand the nuances of the field, can recognize the "unknown unknowns", and eventually have the ability to push the boundaries of existing knowledge. Maybe that is what we should really call learning: not simply acquiring information, but developing enough understanding and practical experience to contribute something new to the field. Ironically, on the same front page of HN, there is a post about Andrew Wiles and this. I don’t think I would be able to comprehend Fermat’s Last Theorem, the Poincaré Conjecture, or Gödel’s Incompleteness Theorems, even with the availability of LLMs.