Bootstrap

John Markoff’s piece about IBM’s Watson contrasts John McCarthy and Artificial Intelligence with Doug Engelbart and Intelligence Augmentation. This is an interesting and subtle argument.

The first conference event I co-chaired was the 1988 AAAI Workshop on AI and Hypertext. We opened with Ted Nelson on Xanadu: The One True System, and followed with Doug Lenat on Cyc: The One True Ontology. I looked at Cyc as a wonderful extension on hypertext – remember, in 1988 both Cyc and the Web seemed far, far in the future – but Lenat saw them as natural competitors. His concern, as I recall, was that if hypertext turned out to be good enough, it would set real machine understanding back for a generation while people tinkered with links and retrieval.

That’s pretty much what Watson does, as I understand it: it mines lots of hypertextual information to find likely answers, while not trying to build much representational depth. I haven’t read the papers; I may be wrong.

It would have been easy for Markoff, writing under the absurd headline “A Fight to Win the Future: Computers vs. Humans,” to make Engelbart the tool-building hero, giving people better tools instead of automating their jobs in opposition to the scary AI robot-builders. (Do androids dream of LED sheep?) But Markoff also captures the false-sounding note in the Augment doctrine:

Also that year the computer scientist Douglas Engelbart formed what would become the Augmentation Research Center to pursue a radically different goal — designing a computing system that would instead ‘bootstrap’ the human intelligence of small groups of scientists and engineers.

Small groups of scientists and engineers. That’s not unfair; I’ve heard Engelbart describe the idea in talks over two decades and more, and that what Doug sounds like. I’ve always winced. But take away the qualifying clause and we’re fine, the dream is intact, all is well. It might begin with a few people; not everyone learned to read at once, not everyone discovered painting at the same time. But I think most of us have always been working for everyone, not for small groups.