Before the 2000s-era machine learning revolution, AI was assumed to be solvable via symbolic reasoning. Pure logic, in other words. See: Cyc. Cyc was/is an attempt to encode all of human knowledge using predicate logic. If you have knowledge in this form then you can use the standard rules of logic to infer new knowledge and do reasoning.

TBL/W3C felt like the next big upgrade for humanity would be AI that could understand the web. To achieve that, you need web pages to be encoded in the form of logical triples. Cyc used a custom lisp-ish language called CycL to do that, RDF was the same concept but 'web-ified' i.e. using XML and URIs for everything.

Of course that approach never took off. Even by 2003 Google was starting to master machine learning techniques like scalable logistic regression, the webbiest of web companies just didn't care about this symbolic AI approach at all because in reality nobody was using it. Too complicated, too abstract, no tools or cool demos. Pure vapourware in other words. GPT-3 has now shown that you don't need everyone to learn new syntaxes or predicate logic to train AI on the web. You can rely purely on statistical techniques and neural networks.

This is fantastic historical information. It's even more fascinating(and horrifying) to think that these things collided the way they did and derailed the evolution of a maintaining a proper HTML spec and that ultimately one company - Google shifted both the direction of AI movement as well as the direction of the browser. Thanks for the wonderfully detailed responses, I really appreciate it.

You're welcome!