Sep 30, 2026 · 31 min · 11 segments
Why Symbolic AI Failed — and Then Won Become an Omega or Aleph Member on Yotuube and help us keep this going! 🙏 [https://www.youtube.com/@aboutlogic](mailto:https://www.youtube.com/@aboutlogic) Or…
Thorsten AltenkirchHostDeniz SarikayaHostI mean, I, I learned Lisp and Prologue and, and, and I, I was, I was really, uh, excited about this.
Uh, there was, uh, lots of investment and first generation computing and stuff like this, yeah? And it all, let's say it came, it came to, to more or less nothing.
Uh, well, also expert systems were created, and I was working for companies, uh, who, who, who, who, who worked in this, in this area.
And more and more, I, I started to have, uh, a bad feeling because I realized that the promises which were made were not met with actual, uh, uh, tools or, or programs or systems which, which would fulfill these promises.
I was feeling a little bit guilty a-about being involved in, in, in this, in this, in, in, in this, uh, overblown, uh, prom-promises, which is also the reason why I personally moved away from AI and went more into formal logic, uh, uh, theory of, uh, programming languages, lambda calculus, and so on.
Because this was a bit so- this was solid, yeah? Whereas this AI stuff w-was, uh, was, was like a, a clearly a bubble.
And, and, uh, the s- the idea of, of the old AI was to, was to base, uh, artificial intelligence on symbolic structures, on symbolic logic, on, yeah, epistemic logics, uh, uh, uh, logics which take into account how moral, how people actually reason, logics of, of default and modality and so on.
This was all put in, uh, uh, and there were grammars which were much more evolved than the sort of standard Chomsky s- style, style grammars, but had much more complicated structures to model, uh, actual human languages.
And, and, uh, tools were based, were built based on these, uh, grammar-based language models.
And it was quite depressing to see, uh, how they performed, because as long as you stuck to the, uh, sort of script, as long as you sort of knew what the system could swallow, you could communicate, and it would look like a, a normal sort of, uh, interaction.
But if you put somebody who didn't know, uh, how to talk to the system in front of it, it, it, it, it would completely fail.
It was very brittle, yeah? So, uh, that was the situation, and as I said, as personally for me, I mean, uh, uh, it, it had the consequence that, uh, even though I had some jobs to do with AI, uh, in the end, uh, for my academic career, I, I decided that I didn't want to pursue this, but instead do, uh, stuff which I'm still doing, type theory, lambda calculus, logic, and so on.
I mean, I, I learned Lisp and Prologue and, and, and I, I was, I was really, uh, excited about this.
Uh, there was, uh, lots of investment and first generation computing and stuff like this, yeah? And it all, let's say it came, it came to, to more or less nothing.
Uh, well, also expert systems were created, and I was working for companies, uh, who, who, who, who, who worked in this, in this area.
And more and more, I, I started to have, uh, a bad feeling because I realized that the promises which were made were not met with actual, uh, uh, tools or, or programs or systems which, which would fulfill these promises.
I was feeling a little bit guilty a-about being involved in, in, in this, in this, in, in, in this, uh, overblown, uh, prom-promises, which is also the reason why I personally moved away from AI and went more into formal logic, uh, uh, theory of, uh, programming languages, lambda calculus, and so on.
Because this was a bit so- this was solid, yeah? Whereas this AI stuff w-was, uh, was, was like a, a clearly a bubble.
And, and, uh, the s- the idea of, of the old AI was to, was to base, uh, artificial intelligence on symbolic structures, on symbolic logic, on, yeah, epistemic logics, uh, uh, uh, logics which take into account how moral, how people actually reason, logics of, of default and modality and so on.
This was all put in, uh, uh, and there were grammars which were much more evolved than the sort of standard Chomsky s- style, style grammars, but had much more complicated structures to model, uh, actual human languages.
And, and, uh, tools were based, were built based on these, uh, grammar-based language models.
And it was quite depressing to see, uh, how they performed, because as long as you stuck to the, uh, sort of script, as long as you sort of knew what the system could swallow, you could communicate, and it would look like a, a normal sort of, uh, interaction.
But if you put somebody who didn't know, uh, how to talk to the system in front of it, it, it, it, it would completely fail.
It was very brittle, yeah? So, uh, that was the situation, and as I said, as personally for me, I mean, uh, uh, it, it had the consequence that, uh, even though I had some jobs to do with AI, uh, in the end, uh, for my academic career, I, I decided that I didn't want to pursue this, but instead do, uh, stuff which I'm still doing, type theory, lambda calculus, logic, and so on.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
