Symbolic artificial intelligence

They explored other wiring approaches.

But fundamentally, the thing they believed in was emergence, and their main argument was GOFAI failed and we are not GOFAI.

Today you've never heard of them because they miserably failed just like I said they would.

Then there's this completely different other set of people who figured out how to run program search efficiently using first-order gradient information.

Great, and we wanted to talk a little bit about AI because, I mean, we have this sub-series ongoing with a few episodes more focused on this topic.

And I wanted to inquire a little bit about the relation between GOFAI, so good old-fashioned AI, and this more statistically based, if you want to phrase it like that, these large language models where we have these problems of hallucinations and stuff like that.

But of course, on the other side, which seems to be the much more potent technology in some sense.

Um, so you sometimes say something like the GOFAI approach seems failed.

Uh, why do you think so? Or-

Okay, so I, I'm, I'm a child of, of, of GOFAI.

Uh, when I was a student, a undergraduate student at TU Berlin in the 1980s, uh, I was really excited about AI.

And I went, uh, to the German Spring School on AI, uh, KI Frühjahrschule, and to the German workshop on AI even when I was a student.

... which was in the '80s, right? Uh, yeah, it was in the '80s? Yeah, it was in the '80s.

Uh, I was very fascinated by AI, uh, and this was what we now call GOFAI.

It's good old-fashioned AI, right? And this was structure-based, yeah? It was like-

Mm.

Mm.

Yeah, yeah.

And it seems to me that category theory, for example, and type theory, they are b- better, uh, they fit better with this GOFAI picture, yeah? That, that, that, because they are, they're very good in, in, in explaining structure.

But what I wonder and what you're doing is, h- do they really fit so well with the-

Yeah.

And was a very early criticism of artificial intelligence or what, um, uh, John Haugeland called the GOFAI, the good old-fashioned artificial intelligence, which was very different from the current, uh, large language model statistical, uh-

Yeah, yeah

... structure.