zlacker

> Lenat personally did not release the source code of his PhD project or EURISKO, remained unimpressed with open source, and disliked academia as much as academia disliked him. Most open information concerning Cyc had been deliberately removed circa 2015, at the moment when Cycorp pivoted to commercial applications.

Makes one wonder how much a research being open makes a difference in its real-world success in the current age. Cyc's competitors (LLMs etc.) arguably have a lot to attribute to open public participation for its successes. Perhaps things would have been difference had Lenat been more open with the project?

>>dartha+(OP)
Probably not, tbh. The issue with Cyc is that it required huge amounts of manual effort to create the rules, while LLMs can learn their own rules from raw data.

There was no machine intelligence in Cyc, just human intelligence.

>>Legend+B
So more of an expert system?

>>Legend+B
>while LLMs can learn their own rules from raw data

Supervised vs. unsupervised, but LLMs haven't made any new discoveries on their own ... yet.

>>dartha+(OP)
> Lenat personally did not release the source code of his PhD project or EURISKO...<

Now that Lenat is dead, can his PhD project code and EURISKO code be released?

>>lenerd+y8
Cyc was exactly an expert system (and those were exactly as central an "AI" technology as LLMs are today, a few rounds of AI hype ago.)

>>lenerd+y8
Very much so.

>>giardi+En
Source code for Eurisko: github.com/seveno4/EURISKO

Source code for AM: https://github.com/white-flame/am

>>Legend+B
> There was no machine intelligence in Cyc, just human intelligence.

This can be rephrased for other "AI" projects with exactly the same relevance:

There is no machine intelligence in LLM bots, just human intelligence.