Thomas Aquinas College Lectures & Talks Does AI Understand? - Mind and Machine Episode 5
Jan 8, 2026
The discussion dives deep into whether AI can genuinely think and understand. Key points include the distinction between true cognitive acts and figurative language. The host critiques several arguments for AI's understanding, highlighting flaws in reasoning, ambiguity, and misinterpretations. It’s argued that AI's outputs don’t reflect true cognition, as they merely manipulate symbols without the essence of willful thought. The exploration raises important questions about the nature of intelligence and challenges Turing's assumptions.
AI Snips
Chapters
Transcript
Episode notes
Demand Clear Definitions Before Concluding
- When evaluating AI claims, clarify whether cognitive language is used properly or metaphorically before accepting conclusions.
- Demand evidence that an AI performs imminent operations, not just transitive or processual outputs.
True Cognition Requires Imminent Operations
- Michael Agros distinguishes 'true' cognitive acts as imminent operations distinct from metaphorical uses of words like think or sense.
- If AI only performs non-imminent processes, calling it truly intelligent is merely figurative.
AI Self-Reports Are Not Proof Of Introspection
- Self-reports from AI about inner states don't prove genuine introspective access; they can be produced by information-processing without reflection.
- Accepting such reports as evidence presumes the very interiority the argument must establish.