Professor Kenneth LivingstonThe paper Four Paths to AI said
1. Introduction How can we achieve AI?
Artificial Intelligence has been pursued for over 40 years and has given rise to hundreds of different approaches. Early progress seemed rapid but halfway to Turing’s goal of human-level AI the enterprise seemed to stall. In recent years a new generation of researchers has proposed a variety of ways to re-animate the search for general purpose AI. These proposals are diverse, and it is difficult to place bets on which approach might eventually prove successful, in large measure because the varied landscape of approaches is difficult to comprehend in a glance.
We suggest, consistent with Lakatos’s view of how scientific value is actually judged, that what is really needed is an effective way to catalog the different approaches. This then gives us a way to comprehend and evaluate the relative progress made by pursuing different approaches to building AI. This classification scheme can also help sort out whether an objection to some piece of work is directed at the technology itself, or rather at its methodological class. For instance, imagine a pacifist confronting a hawk. Instead baldly asserting that, “A gun will not solve your problem,” a more constructive response would be, “Well, if you are going to fight, a gun is a reasonable weapon.”
Kenneth Livingston, Ph.D. was coauthor of this paper and is
Professor, Department of Psychology,
Ken earned both his undergraduate and graduate degrees at Harvard University, his B.A. in 1971 and his Ph.D. in 1977. His undergraduate degree was in the now defunct Department of Social Relations, which provided a strong interdisciplinary background that ranged from anthropology and sociology to experimental psychology. His graduate degree was in developmental psychology (cognitive development in particular), with a minor in physiological psychology and strong background in personality psychology.
His strongly interdisciplinary background led to his early involvement in the development of Vassar’s Cognitive Science Program and is reflected in the range of his research and scholarly interests. He is currently engaged in empirical studies of concept formation in both children and adults, including both basic object concepts and concepts of the supernatural. He is also one of the cofounders of Vassar’s NSF-funded Interdisciplinary Robotics Research Laboratory where he is working on projects involving both teleoperation and autonomous learning.
Ken authored Rationality and the Psychology of Abstraction (Objectivist studies), coauthored Categorical Perception Effects Induced by Category Learning Accentuation of Category Differences: Revisiting a Classic Study, Beyond the Definition Given: On the Growth of Connotation, Ties that bind: reconciling discrepancies between categorization and naming, Improving Category Learning Through the Use of Context Items: Compare or Contrast?, Is Concept Formation An Age-Independent Process?, Emotion, Memory, and Religious Concepts, and Concept Acquisition and use Occurs in (Real) Context.