In his first chapter about hyperlinks and the organizational structure of information systems, Dreyfus emphasizes what he views as the limitations of the internet as a source of relevant data. He states that we possess commonsense knowledge by having individual thoughts and bodies, and that machines will be incapable of sorting information based on a commonsense they cannot possess. He also describes the current rather imperfect system in place for finding things on the internet. While I think he brings up many good points, and makes the reader consider things normally taken for granted, I found his first chapter very defeatist overall. He laments the lack of any decent system to organize the swarm of information on the web, and ultimately states that no such system is even possible. He criticizes the work of those like Lenat on systems that try to take a first step towards more intelligent computer retrieval systems, even though their success (on even a small level) would accomplish some of the intelligent organization Dreyfus seems to desire.
I think it's fine and well to say whether something is theoretically possible or not, but all Dreyfus seems to say is that our system is lousy, people haven't had too much success with making it better, and it probably won't ever get any better. That doesn't seem like a very constructive or useful approach to me. Granted, if it is indeed impossible to improve the system in any way it will save time by not wasting effort trying to improve it. But I think advances in computer technology and software are still on a fast enough level that it may be possible for people like Lenat to achieve breakthroughs at some point in the future. Personally I don't rule out the possibility of AIs somewhere far in the future, it would certainly be an exciting, if not altogether likely, development.
No comments:
Post a Comment