The Yale artificial intelligence project

An AI lab is like a greenhouse. Researchers develop new ideas and plant them in programs. The programs are cultivated, hybridized, nurtured. The weaker ideas die out. The stronger ideas are grafted onto new stock and serve as the basis of hearty new strains. At Yale, there has been a traditional summer seminar series at which graduate students present their unprepossessing theories to the vocal and critical review of their colleagues. The tenor of these discussions is conveyed by their title: “The Friday Fights.” Such occasions provide the Yale researcher opportunities for both pruning and growth. Cultivation by candor is the standard. This level of peer review has also been experienced by colloquium speakers. Many visitors to the lab were unprepared for the onslaught. By now though, Yale’s reputation for open debate has led many speakers to agree to disagree. Ideas and theories grow through this process of natural selection. The best are represented here, in this collection of technical reports from the past dozen years. These reports are our harvest-the results of twelve intense years of work by a large crop of researchers. In the present article, we survey this collection, which is now made available on microfiche through Scientific Datalink. The work falls into several areas.