Automated Reasoning

The panel will address the following questions:

• Can an AI discover, revise, and reject its own axioms, and the inference schemata, heuristics and implementation that are used to reason over those axioms?

• The word "reasoning" can encompass a wide range of activities for intelligent humans - which of these may have been omitted or neglected in AI in the past, but could potentially be subjected to some form of axiomatization and AI reasoning?

• What would it take to do reasoning to "all of science" scale?

• When is formal axiomatization indispensable to precise and broad-ranging reasoning — and when is it a hindrance? To what extent does this depend on the nature of the axiomatization?

• Is an explicit context mechanism, and axiomatization of context, like the Cyc MicroTheory system, necessary for reasoning in AI?