> Date: Sun, 9 Mar 2025 21:14:43 +0300 > From: Jean Louis <bugs@gnu.support> > Cc: Eli Zaretskii <e...@gnu.org>, van...@sdf.org, > emacs-tangents@gnu.org > > > Alpha Go changed fundamentally how humans play Go. > > > > Here’s a nice to read discussion of a hobbyist’s understanding of > > insights Alpha Go provided for LLMs: > > https://www.lesswrong.com/posts/WW7bvpZmgBihhoJPk/planning-in-llms-insights-from-alphago > > > > Based on that it is clear that there are problems we can currently only > > solve with LLMs. > > Okay, let's go back to previous known facts: > > - There are problems we can currently only solve with computer. > > We can inspect that statement: > > - There are problems we can currently only solve with shell tools. > - There are problems we can currently only solve with statistical programs. > - There are problems we can currently only solve with C language. > - There are problems we can currently only solve with LLM. > > All those problems are computing programs. > > I am very sure that computers are really made for combinatorial-based > computations. > > But is the statement something really new?
You evidently have a very restricted notion of what is "new", which seems to be specifically tailored to your goal of asserting that LLMs and ML technology in general cannot produce anything that is traditionally considered to be result of creative activities. You basically fire first, and draw the target around the hit point later. It is pointless to argue with you as long as this is your attitude, because you are not open to considering ideas and opinions different from yours. (And no, ML and LLMs do not engage in combinatorial computations, they solve problems differently, and that is why they succeed, because in many cases the problems are NP-hard, so cannot be solved by dumb exhaustive search of the solution space.) --- via emacs-tangents mailing list (https://lists.gnu.org/mailman/listinfo/emacs-tangents)
