Should We Build Conscious Machines?

They debate ethical, legal and practical reasons to build consciousness, including responsibility and potential suffering.

Play episode from 01:06:24

chevron_right

Transcript

chevron_right

Transcript

Episode notes

Dr. Michael Timothy Bennett is an award-winning AI researcher at the Australian National University and author of "How to Build Conscious Machines" — a mathematical blueprint for why anything is conscious and what it would take to engineer a machine that truly experiences the world.

Expect to learn why information processing feels like anything at all, how consciousness is built from a hierarchy of selves, why a fly has experience but a slime mold does not, why current AI is fundamentally incapable of consciousness, how cancer is an alignment problem between cells, what building a conscious machine actually requires, and much more

…Timestamps:

00:00 Why Does Information Processing Feel Like Anything?

01:05 The Cosmic Ord: How the Universe Deletes What Doesn't Work

03:59 Where is the Line Between Non-Conscious and Conscious?

05:53 Multi-Scale Competency: Cells, Organs, Organisms, Ecosystems

07:43 The First Order Self: Phenomenal Consciousness Explained

08:27 Access Consciousness: Predicting Your Prediction of Me

13:04 When Do Enough Abstraction Layers Produce Experience?

18:40 Liquid Brains vs Solid Brains: Ant Colonies and Human Networks

21:03 Bioelectricity: How Cells Signal and Cooperate

24:03 Human-AI Hybrids: Prosthetic, Cooperative and Integrated

30:30 Is There a Bottom to the Abstraction Stack?

37:07 Why a Single Thread CPU Will Never Feel Anything

40:56 Feeling Pain vs Knowing You Feel Pain

42:52 Do Dogs Model Your Mind?

46:58 The Paper He Wrote as a Joke That Won Best Student Prize