Monday, July 24, 2006

Artificial Intelligence vs. Intelligence Augmentation

This is antother subject I've been thinking for a while until I came across other people wondering about it and formallizing it as well.

The simplest way to think about it is, what path of evolution will win?
  • making machines think and become intelligent? or,
  • making machines just power the bionic brain?

  • In the first case, we're talking about "Artificial Intelligence", which has been on and off a hot thing in Information Systems throughout the times. It's like it'd be nice to have, and it's doable as well, actually we've been adding some kind of AI to our systems all this time. But ... we don't seem to be ready to make machines learn the way we do, to think the way we think, to interact the way we do.

    In the latter case, we're talking about what has been called "Intelligence Augmentation" or "The Bionic Brain". I came across some of Tim O'Reilly's posts mentioning these denominations. This leaded me to Pattie Maes, in her article "Intelligence Augmentation, a talk with Pattie Maes". She studied for years with the AI guru and one of my personal heroes Marvin Minsky (first ever mentioned to me by Nicolas Jodal when I asked him some weird questions about computer learning many many years ago). Back to the IA thing, what seems to be the alternative to making machines think, is to use computer networks, software and databases as the wires of our brain, as the myelin that keeps humanity super brain together, glued, connected. The internet and all the cool software we're building today, making users collaborate, making wired social networks, being a live memory for society (like having as global mind a bunch of extra RAM memory).

    It's hard to tell which of these evolutionary paths will win ... I would think that a combination of both, machines will become more and more intelligent and the bionic brain will grow even more so because of the addition of machine intelligence to human intelligence. Maybe the future is collaboration between machines and humans? humans teaching computers their strengths while using machine strengths for their own empowerment and the oposite hopefully as well. I for one will be some day writing manifests about machine "human????" rights :)
    Anyway, it all will come back to cyborgs in the end, that's my humble opinion.

    In AI it looks like Bayesian belief networks is a promising thing (thanks Jodal for this one too). In IA topic maps and semantic web seem to be the future.


    You Mon Tsang said...

    I think IA is the evolutionary path in the short term and AI may be the path in the long term.

    I've been in the software business for a while now so perhaps I am hampered by thinking about applied technology (where AI seems like an expensive pipe dream).

    In the short term, IA means that we can program machines to do the work that humans do well but do not add a ton of value (audio transcription, tagging and categorization, photo searching, making connections in a very large network, etc). That seems most do-able and useful.

    However, I do expect progress made in both realms in the near future. IA will be behind the scenes, less flashy, but more useful in the near term.

    Anonymous said...

    Here are some links that I believe will be interested