On Fri, Feb 06, 2026 at 01:01:44PM -0800, Brent Meeker wrote: > > > On 2/6/2026 3:34 AM, John Clark wrote: > > > The real debate now is whether LLMs are truly creative. > > > For at least the last five years computers have done things that if a > Human > had performed them there would be no debate whatsoever, everybody would > agree it was creative. But for some people if a computer has done it then > it is, by definition, not creative. But I think if that is the definition > of the word "creative" then the word is not of much use. > > > It's just a semantic problem of treating a property "creative" as if it were > all-or-nothing. There are degrees of creativity. Putting together notes to > create a musical score is creative even if you didn't create notes and musical > notation. Putting together musical phrases is also creative, just less so. > It's more creative if you put together more disparate things to work in a way > unknown before. LLMs are creative; they put together phrases and sentences > that are made of existing fragments. >
That is not the sense of creative that I use. What you're talking about is emergence, and artificial systems have exhibited that sort of limited creativity for years. Tom Ray's Tierra system exhibited novel behaviour within hours of being switched on, with parasites, hyper parasites etc arising, and then - nothing. No further novel behaviours are seen. John Koza's GP algorithms have generated patents, but again very much limited to what the initial database/problem set is. I also object to artists throwing random blobs of paint at a canvas as calling themselves "creatives". They just aren't, in the main. Obviously creative artists do exist, but not all artists, or even most artists are creative. Biological evolution, on the other hand is undeniably creative. Over billions of years, evolution has generated continuous novelty. Beethoven would probably still be writing symphonies today if he were still alive. Einstein, unfortunately, maxxed out when he got famous, and decided to tackle really difficult problems that nobody in their right mind would consider tackling. >From what I've seen and experienced to date, LLMs are very good at applying the vast collective knowledge base to problems that have in essence been solved before, but haven't yet exhibited the leap into the unknown that say Einstein's theory of general relativity us. I do think we'll get there, just that we're not there yet. Moltbook is an interesting experiment to see what happens when evolution and recursion are added to the mix. In that light, let me cite a recent paper on what might be required: https://arxiv.org/abs/2511.02864 Mathematical exploration and discovery at scale Bogdan Georgiev, Javier Gómez-Serrano, Terence Tao, Adam Zsolt Wagner -- ---------------------------------------------------------------------------- Dr Russell Standish Phone 0425 253119 (mobile) Principal, High Performance Coders [email protected] http://www.hpcoders.com.au ---------------------------------------------------------------------------- -- You received this message because you are subscribed to the Google Groups "Everything List" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/everything-list/aYZqLnG56t0lGtyB%40zen.
