Insiders' Critiques of AI Methods

AI researchers tend to be gung-ho people, happier trying to achieve something than worrying about the hows and whys. But sometimes it is worthwhile to step back and take perspective. This page lists some discussions of the limitations of AI approaches, mostly explanations by former AI researchers of why they left the field.

AI approaches do not always work well

• Artificial Intelligence Meets Natural Stupidity. by Drew McDermott.
  in Mind Design: Philosophy, Psychology, Artificial Intelligence, J. Haugeland, ed., pp. 143-160, MIT Press, 1981.
  common pitfalls in AI research
• Intelligence Without Reason by Rodney A. Brooks.
  in Proceedings of the Twelfth International Joint Conference on Artificial Intelligence, pp. 569--595, 1991.
  historical perspective, AI methodology, personal history
• Toward a Critical Technical Practice: Lessons Learned in Trying to Reform AI, by Philip E. Agre.
  in Bridging the Great Divide: Social Science, Technical Systems, and Cooperative Work. Geof Bowker, Les Gasser, Leigh Star, and Bill Turner, Eds. Erlbaum, 1997.
  intellectual history of AI, some problems with AI methodology, personal history
• Artificial Intelligence and Other Approaches to Speech Understanding: A case study in methodology. by Nigel Ward.
  Journal of Experimental and Theoretical Artificial Intelligence, 10, 487-493, 1998. (abstract),
  point-by-point comparison of the AI approach with scientific and engineering approaches

AI visions are not necessarily a good inspiration for computer systems designers

• Agents of Alienation by Jaron Lanier.
  Interactions 1995, volume 2, 66-72.
  discussion of problems with the vision of intelligent agents for user interfaces
• Shifting Viewpoints: Artificial intelligence and human-computer interaction. by Terry Winograd.
  Artificial Intelligence 170(2006) pp 1256-1258.
• It's Better to be a Good Machine than a Bad Person: Speech Recognition and other Exotic User Interfaces at the Twilight of the Jetsonian Age. by Bruce Balentine, ICMI Press 2007.
  essays on spoken dialog systems, personal history

AI techniques are not always a harmless thing to teach undergraduates

• Programming as a Discipline? by Nigel Ward.
  Communications of the ACM, 40(12), December 1997, pp 113--114.
  effects of courses in "AI programming" and "programming paradigms", personal history

Also see the AI Critics page.