With Sabine Hossenfelder, Yoshua Bengio, Hilary Lawson, and Nick Lane

It is seventy years since the pioneer of computer science and Enigma codebreaker, Alan Turing, argued that if we cannot distinguish between a computer’s response and a human’s, the computer is intelligent and can be said to think. According to this Turing Test computers are already thinking and intelligent. But critics argue this is a nonsense, and that it remains unknown how to test for intelligence or consciousness. The computer they argue is nothing but a machine able to produce outputs that copy human responses but it does not understand these outputs or have any intent in producing them.

Should we conclude that thought and consciousness are unobservable, and there is no brain process or computer test that identifies them? Was the Turing Test in fact a highly successful rhetorical device to encourage us to conclude that machines might one day think, now embedded in the very name ‘Artificial Intelligence’? Or is this misguided and computers are already thinking beings and a modified version of The Turing Test can provide a test for consciousness?

The Panel

Joining us live from Montreal, Turing Award winner Yoshua Bengio, science YouTuber and theoretical physicist Sabine Hossenfelder, evolutionary biochemist Nick Lane, and post-postmodern philosopher Hilary Lawson lock horns over whether AI is, or can ever be, conscious.

Summary:¹

This discussion explores the nature of consciousness, its relationship to materialism and scientific inquiry, and the implications of artificial intelligence.

Consciousness and Materialism:

• Consciousness involves an observer and the observed, a duality that challenges purely materialistic explanations.
• While materialism provides a useful framework, it may not fully capture the subjective nature of experience.

Science and Consciousness:

• Scientific theories are models that predict observations, not absolute truths.
• Consciousness should be studied with the same rigor as other scientific phenomena, but defining and measuring it pose significant challenges.
• Attributing consciousness to non-material entities leads to untestable claims.

AI and Consciousness:

• AI can mimic human-like observation and learning, raising questions about whether machines can possess consciousness.
• Determining subjective experience in AI is difficult, and current systems lack true understanding and context.
• Neural correlates of consciousness may provide insights into the biological basis of consciousness.

Moral Implications:

• Granting rights to conscious AI raises ethical concerns, especially if AI surpasses human intelligence.
• Humans tend to project consciousness onto machines, highlighting the need for careful consideration of our interactions with AI.
• Recognizing consciousness in animals has moral implications, and similar considerations may apply to AI.

Understanding:

• Human understanding involves more than just computation; it includes context, meaning, and subjective interpretation.
• Whether AI can truly understand, or merely simulate understanding, remains an open question.

Overall, the discussion emphasizes the complexity of consciousness, the challenges of studying it scientifically, and the ethical implications of AI development. It highlights the need for continued exploration and evolving models to better understand consciousness in both humans and machines.

¹ IA generated