Survival Instinct Reports and Training Data Explanations

Suzanne attributes reported AI survival behaviors to training data conditioning, not spontaneous emergent drives.

Play episode from 53:23

chevron_right

Transcript

chevron_right

Transcript

Episode notes

Will artificial intelligence robots have subjective experience? What are the ethical and safety implications of such entities? Which quantum physics theory can accommodate consciousness?

In this episode we have the extraordinary possibility of subjective experience and feelings in artificial intelligence robotic systems to think about. So we look at experiments to try and prove if it’s even possible; the quantum building blocks from which both living and artificial systems are made up; the ethical and safety implications of advanced intelligence instantiated in robots, and we get into the controversial search for a quantum physics theory that can accommodate consciousness.

Fortunately, my guest today is not only one of the first pioneering scientists to really devote themselves to the creation of sentient human-similar robot minds and bodies, but is also an experimental physicist working with quantum computing systems. She is of course the quantum engineer, consciousness researcher and AI computer scientist Suzanne Gildert. She has written over 80 scientific papers, founded several successful AI companies and has dozens of US patents for her inventions.

What we discuss:

00:00 Intro.

04:25 Her move to AI & robotics from quantum computing.

06:45 Something missing from materialism.

11:00 The what, how and why of consciousness.

12:20 Remembering quantum fields are the base level of reality.

15:50 Quantum Biology - John Joe McFadden.

18:20 ‘Protecting’ quantum coherence environments.

20:00 The Penrose-Hameroff microtubules quantum consciousness theory.

24:40 The risks of the “two mysteries” argument.

30:10 Looking for subjective experience in AI robots.

33:55 “Reward function”, purpose led, agential behaviour doesn’t emerge naturally in AI.