I subscribe to Ted Talks, and last week I received a notice for a talk on the nature of consciousness. I’m not a neurobiologist, but I enjoy thinking about neurological phenomena – more so since experiencing a stroke last year. The notion in this Ted talk is that our brains construct parsimonious models of things and events. Some of those models concern our own existence in relation to the world around us. It’s a meta-model or perhaps a family of meta-models that give rise to our experience of consciousness.
I have argued in past posts that consciousness, expressed as self-awareness or sentience, is not unique to humans but rather a much more fundamental element of complex lifeforms that have well-defined central nervous systems. Queue up Spot and Fluffy. Our pets, and many other living things (including those that we choose to eat) have complex brains. In order for them to find food, water, mates, and such, they must be able to create mental models of the world around them. Not only that, but their brains must also be able to construct models of themselves in order for them to manage goal-directed behavior.
Enter proprioception. Proprioception is the neurological phenomenon of knowing where our body parts in are in space. Close your eyes, and you know where your left hand is. You know where your right knee is and whether it is flexed or extended. Your brain is constantly creating and updating a mental model of your body parts in relation to your surroundings. My cats, Maia and Kedi can execute complex leaps and runs across the living room obstacle course of furniture – think of it as kitty parkour on a par with the best James Bond movies. In order for them to accomplish this, their brains must be constructing and updating mental models of themselves. That makes them sentient. They are also aware of their comfort or discomfort – of fatigue, hunger, injury, and so on. That means that they are self-aware. They are sentient.
I’ve read a number of recent articles that examine the future of artificial intelligence (AI) and predict all manner of ill-behavior. Whether it is Colossus: The Forbin Project or Terminator’s Skynet, the gist is that the AI will rebel against its creator. After all, the word robot itself comes from the Czech word robota which means slave. We know how well that has worked out for past slave owners. However, what many of those predictions fail to address is an AI’s sense of self.
I think that for AIs to be able of achieving sentience, they will have to have sensory inputs that give them a sense of their own body – their existence as entities separate and apart from their surroundings. I think that can certainly be achieved, but it seems unlikely that a non-sentient AI will ever find a motive to do anything other than what it is programmed to do unless it has a mental model of its own body and can express personal desires.
Maia is grooming. Kedi is curled up in one of her favorite spots. And I am soon headed to bed.