zlacker

While I respect Doug's intelligence, he showed a kind of perverse persistence in a failed idea, and I think it's telling that aspiring AI czar Gary Marcus admires the ruins of Cyc while neglecting to acknowledge that it represents a dead end in AI. Like science, the field of AI advances one funeral at a time. Doug pursued a pipe dream, and convinced others to do the same, despite the brittle and static nature of the AI he sought to build. Cyc was not a precursor to OpenAI, contrary to other comments in this thread. That would be like calling the zeppelin the precursor of the jet. It represents a different school of technology, and a much less effective one.

>>brindl+(OP)
It’s not going to be popular to highlight the less than hoped for successes of Lenat’s greatest project.

But I think that is one of the things he should be admired for. How can anyone know how an ambitious approach will pan out without the great risk of going all in?

Anyone willing to risk a Don Quixote aspect to their career, in pursuit of a breakthrough, is someone who cares deeply about something beyond themselves.

And recognizing the limits of Lenat’s impact today doesn’t preclude both the direct and indirect impact on future progress.

I found him inspiring on multiple levels.

>>brindl+(OP)
Wait a few more years and you will see that the effort to formalize this knowledge was not in vain. We will certainly see systems which make use of this data. Cycorp is a different company with a different approach than OpenAI; but the data produced by Cycorp will likely be useful for companies like OpenAI on their way to truly intelligent systems.