Organoids, LLMs, and tests for AI consciousness, with Prof Susan Schneider

Written By Will Millership

In the second episode of Exploring Machine Consciousness, we welcomed philosopher and AI expert Professor Susan Schneider to discuss consciousness, quantum physics, and the future of intelligent machines.

Superpsychism: consciousness at the quantum level

Susan introduced her theory called "Superpsychism," proposing consciousness arises from quantum coherence and entanglement at the fundamental level beneath spacetime. She identifies with a physics-based variant of panpsychism, suggesting a deeply entangled quantum level is inherently more conscious than traditionally recognized physical substrates.

Can machines become conscious?

Susan believes it is possible for machines to develop consciousness; however, she is sceptical that current silicon-based technologies, particularly GPUs, are going in that direction. She sees consciousness as more likely to arise from technologies such as organoid computing, lab-grown neural structures. "When you start to grow larger ones, you start to see patterns that resemble the brain... I think that technology is growing in sophistication."

The problem with large language models (LLMs)

Despite impressive performance and convincing interactions, Schneider argues that today's popular AI models like ChatGPT or Claude are not genuinely conscious. Instead, she describes their behavior as "conceptual mimicry," derived from a vast amount of human-generated data. She refers to this as the "crowdsourced neocortex theory," emphasizing: "their avowals of consciousness do not say anything about whether or not they're actually conscious."

Hybrid systems: the future of machine consciousness?

Susan sees consciousness as most likely to arise in hybrid systems combining biological technologies like organoids with digital technologies such as LLMs. "It'd be really interesting to construct an artificial system through organoids and then hook it up to an LLM and have the LLM report its experience."

A thoughtful path forward

While wary of the near-term pitfalls surrounding the development and misunderstanding of machine consciousness, Schneider is hopeful for a responsibly managed future: "If we get past this inflection point, I think that it's going to be incredibly exciting." She thinks that advances in hybrid systems could unlock profound opportunities, from medical breakthroughs to new understandings of consciousness itself.

Links:

Will Millership
Previous

The LaMDA Moment: What We Learned About AI Sentience
Next

Anticipating an Einstein moment in the understanding of consciousness, podcast with Henry Shevlin