zlacker

"Their topmost distinction was between things with souls and things without souls. And large trees were in the former category, whereas small trees were in the latter category…"

This reminds me deeply of Borges: https://en.wikipedia.org/wiki/Celestial_Emporium_of_Benevole...

To me, that bit of Borges is a reminder that all human taxonomies are limited and provisional. But it seems to me that Cyc and its brethren are built around the notion that a universal taxonomy is important and achievable. I guess it's possible that a useful kind of cognition could happen that way, but it's patently not how people work. If I had gotten to the point where I was forced to define exactly when a tree got a soul, I hope I'd realize that I was barking up the wrong tree.

>>wpietr+(OP)
Lenat in his talks emphasised the infeasibility of building a universal ontology. From what I understand later Cyc modus operandi was a bunch of self consistent micro-theories but not a one huge theory of everything.

>>varjag+7f
Yes, but I'm not sure he fully took in the lesson. Did they really build a bunch of separate domain-specific ontologies? Or, given that it's one system, did they build a universal ontology that's just a big mess?

The lesson I take from the fact that universal ontologies are untenable is that human cognition isn't driven by ontology, so the quest to make a thinking thing out of pile of symbolic logic is one that has no guarantee of succeeding. I think Cyc's whole project is roughly similar to the Frankensteinian notion that if you just put together the right parts and provide a vital spark, you'll get a living being. It might work and it might not, but either way it's not science; it's sympathetic magic with the trapping of science.

>>wpietr+d91
Well that's how LLMs got going: a critical mass of context until it ignited from its own gravitational pull.

And yes Cyc ontology wasn't consistent. Lenat's point was that it is impossible to have an ontology consistent. Which makes sense given how there is no consistence across human society or even every individual human.

>>varjag+BQ3
I wouldn't say that LLMs have "ignited". I think they've gotten steadily better at autocomplete, but I don't think they've crossed any important threshold on the way to the sort of autonomous intelligence that Lenat was after.

I also don't think it was Lenat's original point to have an inconsistent ontology, as evidenced by his early projects. I agree that he eventually had to admit that. What I'm saying is that when he admitted that he should have recognized that it cut at the heart of what he was up to. Something I think borne out by the fact that he spent his whole life on something that didn't succeed on its own terms.