Discussion with Murray Shanahan 2

Summary

The conversation delves into the concept of embodiment, exploring how both biological systems and AI navigate various 'spaces,' from metabolic and transcriptional to 3D spaces. The speakers challenge the traditional notion of disembodiment, suggesting that intelligence and navigation can exist in non-physical spaces. They discuss the limitations of current language and frameworks, emphasizing the need to evolve our conceptual tools to better understand both living systems and artificial intelligence. The dialogue explores the intersection of biology, philosophy, and AI, aiming to bridge the gap between the familiar and the unknown.

Takeaways

• 😀 Embodiment is not limited to 3D space; earlier biological processes navigated through spaces like metabolic, transcriptional, and physiological states.
• 😀 The concept of disembodiment is problematic when applied to organoids or AI, as these systems may embody different kinds of spaces that we don't fully understand.
• 😀 Humans often fixate on 3D space and movement as embodiment, but intelligence can operate across various types of spaces, beyond just the physical.
• 😀 Language models and AI should not be seen as disembodied; their processes may be navigated through non-physical or 'semi-alien' spaces.
• 😀 The conversation highlights the importance of new conceptual tools to understand and accommodate AI and non-human intelligences.
• 😀 Language and philosophical categories need to evolve alongside scientific advancements, especially when conceptualizing living or artificial systems.
• 😀 The discussion focuses on navigating the boundary between the familiar and the unfamiliar, particularly in the context of embodiment and intelligence.
• 😀 Cells and tissues are used as a stepping stone to explore more complex concepts like AI or alien intelligences, which are not easily classified in traditional terms.
• 😀 There is a cultural and social agenda to figure out how to integrate and understand strange, mind-like entities such as AI.
• 😀 Philosophy must be flexible and adaptable, pushing boundaries while acknowledging that we are still participants in the evolving conversation on consciousness and intelligence.

Q & A

What is the main concept discussed in relation to disembodiment?

The main concept discussed in relation to disembodiment is how the term is often used to describe things that don't have a physical, 3D body, but there is a critique that this is too narrow. The conversation suggests that embodiment can take many forms, including abstract spaces like metabolic, transcriptional, and physiological spaces, which are navigated intelligently by living systems, challenging the traditional understanding of embodiment.

How do the speakers define embodiment beyond the 3D space?

The speakers broaden the concept of embodiment by suggesting that living systems navigate a variety of abstract spaces beyond 3D physical space. These include spaces like metabolic, transcriptional, and physiological states, which are harder to visualize but still play a crucial role in how biological systems function and make decisions.

What does the speaker mean by 'navigation' in the context of the conversation?

The term 'navigation' refers to the process of understanding and interacting with different kinds of spaces, whether physical or abstract. This involves determining one's current position, understanding possible movement in that space, sensing states, encountering barriers, and interacting with agents within that space, which applies not only to biological systems but also to artificial systems like AI.

Why does the speaker argue against calling organoids 'disembodied'?

The speaker argues that calling organoids 'disembodied' is misleading because these systems are navigating complex, non-physical spaces that we may not fully understand. Instead of being disembodied, they are engaging with spaces like transcriptional or physiological space, which are very real and important to understanding their function.

How does the speaker's work relate to the boundaries between the familiar and the unfamiliar?

The speaker’s work focuses on exploring the boundary between the familiar (things we already understand, like 3D space) and the unfamiliar (abstract spaces or systems that challenge our current conceptual framework). They aim to bridge this gap by gradually introducing new concepts without leaping too far into unknown territory, especially when talking about AI and new forms of intelligence.

What role does language play in shaping our understanding of embodiment and intelligence?

Language plays a crucial role in shaping how we conceptualize embodiment and intelligence. The speakers argue that our current language is constrained by our familiarity with 3D spaces, and this limits our ability to discuss intelligence in more abstract or non-human forms, such as AI or cellular systems. New conceptual tools and language are needed to address these new forms of intelligence.

How does the conversation address the challenge of defining intelligence in AI?

The conversation suggests that understanding intelligence in AI (and other non-human systems) requires a deeper understanding of how intelligence operates in biological systems, which may not always conform to our traditional, human-centric definitions of intelligence. The speakers propose using concepts like 'navigation' in abstract spaces to better understand how both living systems and AI might exhibit intelligence.

What is the importance of metaphors like 'navigation' in discussing complex topics like AI and embodiment?

Metaphors like 'navigation' help bridge the gap between abstract concepts and more familiar ideas, making it easier to discuss complex topics like AI and embodiment. The use of such metaphors allows for clearer communication and provides a conceptual framework to explore unfamiliar territories of intelligence, whether biological or artificial.

What is the speaker's stance on the potential of artificial intelligence to exhibit intelligence like humans?

The speaker seems cautious about attributing human-like intelligence to AI. They emphasize the need for a more gradual approach in understanding AI's intelligence by first examining how biological systems navigate and process information in non-3D spaces, before making broad claims about AI's capabilities.

What is the 'bridge' the speaker is trying to build in relation to cells and AI?

The speaker is trying to build a conceptual 'bridge' between biological systems and artificial intelligence by starting with cells and their navigation of complex, abstract spaces. This is seen as a stepping stone to understanding how intelligence might operate in AI systems, making the transition from familiar biological concepts to more unfamiliar technological concepts smoother.