Latest advancements in artificial intelligence, such as ChatGPT, naturally lead many of us to wonder – can AI think? Not merely compute in increasingly complex ways and astounding volumes but actually understand what it is doing. Can AI have an experience? And how can a thought experiment devised by Avicenna, an 11th-century philosopher, help us here?

These questions are much more fundamental than they may appear at first sight. They do not simply probe the extent of technological possibilities. What’s at stake here is our understanding of the nature of understanding itself, of what consciousness is and what it means to experience something. Accordingly, our answers will depend on our beliefs about these matters. One such belief, with arguably some of the deepest roots in our modern intellectual heritage, concerns mind-body dualism. 

If my mind in my body is just a ghost in a machine, then it is easy to imagine that “I”, the conscious experiencing person, is the mind that simply resides in this organic body as a type of vessel. If so, then, in principle, any other container will do. Why not an inorganic one? And then, why shouldn’t non-biological entities also be capable of understanding, consciousness, and experience? Such as AI or, as some may prefer to call it, Artificial General Intelligence (AGI), to mark its likeness to human cognition. 

The benchmark of likeness is crucial because, after all, what other reference do we have? How else could we determine whether AI experiences and understands, if not by comparing it to the only point of departure known to us – our own consciousness? This is where we can learn something from Avicenna because, as it turns out, our intuitions about our minds can and do change.

Avicenna’s Flying Man

Approximately 1.000 years ago, polymath Ibn-Sīnā, also known by his Latin name Avicenna, devised the following thought experiment that became known as the Flying Man (FM) argument. 

Avicenna begins by stating that he intends to offer an intuitive way of affirming the existence of a soul (the ‘soul’ here can be understood roughly along the lines of the contemporary concepts of mind, intellect, and reason). This intuitive way calls to mind something already known to us but perhaps in need of reminding. As such, it is a way to knowledge, Avicenna claims, “that will make a powerful impression on someone who has the capacity for noticing the truth by himself, without needing to be instructed, prodded, or turned away from sophistries.”^* We can only hope we have the required capacity and press on. 

Avicenna asks us to imagine the following scenario: “[O]ne of us must suppose that he is created all at once, and created as perfect, but with his sight prevented from seeing anything external [to him]. He is created hovering in the air, or in a void, in such a way that the air does not buffet him so that he would have to feel it. His limbs are separated so that they do not meet or contact one another.”^* In other words, imagine you enter this world as a fully formed, grown-up person with all the usual human capacities but floating in a state where your (completely functioning) senses perceive nothing at all. 

In this state, Avicenna asks you to think if you would confirm your existence (the existence of your ‘self’ or ‘essence’ – translations differ). It is crucial to remember that, in the thought experiment, you have just been created and immediately find yourself in this peculiar state. That means you cannot simply recall what it is like to perceive smells, sounds, temperature, etc. Avicenna’s flying man is created “hovering in a void”. Yet, Avicenna answers his question with a confident conclusion:

“He has no doubt in his affirmation that his essence is existent, even while he does not affirm any extremity among his limbs, nor anything inward among his innards—not his heart or his brain—nor anything external. Rather, he has affirmed his essence while not affirming for it any length, width, or depth. If, in this situation, he were able to imagine a hand or another limb, he would not imagine it as a part of his essence, nor as a condition for his essence… Therefore, as to the essence whose existence he affirms, it is specific for it that it is identical to him and distinct from his body or his limbs, which he has not affirmed.”^*  

_Avicenna

Avicenna is certain that if I am an “alert person”, I will intuitively affirm the existence of my ‘self’ in the absence of any sensory perceptions (not just temporarily but from the first moment of my life). The body and all bodily features are not part of my ‘essence’. Not part of who I am. The body is just a vessel for my ‘soul’ or, in other words, my consciousness, mind, intelligence – the ‘essence’ the flying man has affirmed. This brings us back to an earlier thought: If so, then, in principle, any other container will do. Why not an inorganic one?   

Disembodied Human and Artificial Intelligence 

There is one big problem with Avicenna’s Flying Man thought experiment. I struggle to imagine myself coming into being fully formed but deprived of any bodily sensations, and I do not intuitively affirm my existence in such a state. In fact, the whole flying man scenario strikes me as very counter-intuitive. Incidentally, this shows how questionable it is to build philosophical views on the foundations of seemingly universal intuitions (an insight explored in the video on Continental philosophy I shared in last week’s post). The intuitions that form part of our common-sense beliefs today are subject to social, cultural, and historical changes and influences.

In the flying man, Avicenna brackets all embodied senses and experiences and asks whether something would still be left to experience. Even if we grant his assumption that we are constantly self-aware in the sense of being present to ourselves (even in unconscious states), it is not at all obvious why this presence should remain in the absence of any bodily experiences. Contrary to what Avicenna presumes, our experienced self-awareness is an embodied phenomenon because we have never experienced ourselves in a disembodied state. When I use my imagination and try to picture myself as Avicenna’s flying man, I am doing it all as an embodied creature. It follows that the first-person experience of oneself and all related intellectual abilities are parts of the natural development of an embodied being closely interacting with the environment. 

Avicenna’s flying man was intended to teach us that the human mind is the essence of what we call our ‘selves’ and that it is separate from the body. One thousand years later, it can still teach us something valuable. Namely, the exact opposite: 

Human intelligence is not disembodied; it is fundamentally embodied.

_{humanfactor.blog}

If we accept this, the body is no longer just a vessel for the mind. Our intelligence is a much more complex system of embodied experiences and interactions. What does this tell us about the prospects of a thinking, understanding artificial general (human-like) intelligence (AGI)?

One still widespread view misses the modern flying man lesson of embodied intelligence. It appears to think along the lines Avicenna intended – trying to place a thinking, disembodied ghost into a machine. To understand how humans understand, we need to look at much more than just one organ, the brain: “As such, many researchers believe we’ll need new approaches, and more fundamental insight into how the human brain works, before we can build machines that truly think and learn like humans.” (James Fodor, PhD Candidate in Cognitive Neuroscience, The University of Melbourne)

Another more recent approach acknowledges the fundamentally embodied nature of human intelligence but misses the ‘human-like’ assumption built into our efforts to create AGI. We want a disembodied ‘machine’ to think like embodied humans: “if the mysteries of the brain were to be disclosed, an AGI would be feasible… Everything hinges on figuring out, first, how humans learn and think, and the cognitive interaction between our bodies and the context surrounding us.” (Andrés Felipe Barrero, Ph.D. Candidate in Philosophy, Universität Bremen) 

Even if we develop an artificially ‘embodied’ artificial intelligence, it will be an inorganic ‘body’. That does not mean such a non-biologically ‘embodied’ AGI cannot be intelligent. But it does mean that it will not be human-like intelligence because our intelligence is organically embodied. In fact, we might not even recognise or comprehend an inorganically-embodied sort of intelligence, it being so unlike ours. Can we build something utterly unlike us while aiming at human-like intelligence? Well, accidents happen. 

keep exploring!

P.S. Thank you for visiting me here on the humanfactor.blog! If you enjoyed this post and are interested in more philosophical content, I invite you to explore the blog, leave a comment, like, and subscribe to get notified of new posts (twice a week).

_{*Adamson, P., & Benevich, F. (2018). The Thought Experimental Method: Avicenna’s Flying Man Argument. Journal of the American Philosophical Association, 4(2), 147-164.}

_{Image credit: Photo by Brett Jordan on Unsplash}