Thomas Aquinas College Lectures & Talks Does AI Really Think? - The Mind and the Machine: Episode 4
Jan 1, 2026
The discussion kicks off with a challenge: can AI truly think? Michael Ogros delves into the distinctions between metaphorical and literal thinking. He then outlines ten compelling arguments supporting AI's capability to think, from reasoning to recognizing universals. There's fascinating exploration of the Turing Test, the role of humor in understanding, and AI's creative outputs. The conversation even touches on solipsism, questioning if external behavior can imply consciousness in machines. What fascinating points await rebuttal next?
AI Snips
Chapters
Transcript
Episode notes
Metaphorical Vs. Genuine Thinking
- Michael Agros distinguishes metaphorical from genuine attributions of thinking to devices like thermostats or spellcheckers.
- He emphasizes that human thinking involves imminent operations which simple devices lack.
LLM Logic Quiz And Correction
- Michael Agros recounts testing a popular LLM on syllogisms and getting both correct and blatantly wrong answers over time.
- He notes the model later accepted correction and improved on repetition, suggesting learning-like behavior.
Performance Above Humans As Evidence
- Agros presents the argument that surpassing human performance (e.g., chess, SAT, bar exam) supports claiming AI is truly intelligent.
- He uses historical wins like Deep Blue and AlphaGo and GPT-4 test scores to justify the premise.