Artificial intelligence expert George Zarkadakis explores the idea of consciousness, and just what it is that makes us human.
Can data ever describe, or produce, consciousness? And what would it mean for humanity if it could? Photo: Christiaan Colen
Many notable scientists, including Stephen Hawking, believe that in the not too distant future it will be possible to upload one’s consciousness and live forever inside a computer. Indeed, one of the tenets of ‘strong artificial intelligence’ is that human consciousness can be replicated in a computer.
There is an underlying assumption to such claims, an assumption so deeply embedded in our modern, nay digital, world that we seem to take it for granted: that everything is, ultimately, information. This assumption is so powerful that it has jumped from electronic engineering and computer science – its original craddles – and has made inroads into the ultimate bastion of rationality and the experimental method, namely physics. Profound mathematical similarities between thermodynamics and information theory has led many physicists to ponder whether a more suitable description for the universe ought to be one based on information, a concept beyond matter and energy. Just think of the definition of a black hole: a rupture of time-space knowable only by the absence of information.
So what about the mind? A physicalist would probably argue that the mind is the product of the brain, the result of neurobiological processes. But what if we could account for all the connections and all the processes that take place inside a brain and stored this information in a computer system? What if we could capture a thinking brain as a piece of information, a long coded sequence of noughts and ones? The ‘Human Brain Project’ – ironically headquartered on the opposite bank of lake Geneva from where Mary Shelley penned Frankenstein – aims to do exactly that: to painstakingly record the ‘human connectome’ and upload it on a vast farm of computer servers. It’s a huge scientific ambition and an enormous undertaking. Till today we have only managed to decode the connectome of humble C. Elegans, a hapless worm; and have recently uploaded its ‘worm mind’ unto a tiny robot that rolls around wildly, and presumably just like their real thing. Would the uploading of the human connectome produce a similar result: a robot running around in search of meaning? Seeking revenge from its creators, perhaps? Or longing after a digital lover? Or will nothing happen?
Personally, I expect the latter. And not only because I find the idea of reproducing gothic horror in the age of computing nauseatingly banal. It is mostly because if I am truly just information, and if information is immaterial, then what is the purpose of my body? And what is the purpose of all this matter and energy around us? Besides, this immaterial information idea smells too much like the concept of the soul and of the body as an entrapment; for the believer in pure information there is no matter or energy; everything is mind, pure consciousness, a never ending streaming of bits. As a computer engineer I know this not to be true. But do not take my word for it. Go visit a computer farm, and observe the physical aspect of information, the dull rectangular buildings, the endless corridors of humming machines, the complexity of cement, metal and plastic coming together in multifarious combinations, the asphyxiating heat that this multi-billion dollar physical aspect of information generates. This is the body of information which, despite being material to the core, has somehow become invisible.
George Zarkadakis has a PhD in Artificial Intelligence and is the author of ‘In Our Own Image: Will Artificial Intelligence save us or destroy us?’ (Rider Books). He tweets at twitter.com/zarkadakis.
George is one of the many great speakers exploring all that it means to be a human being and a humanist at this year’s BHA Annual Conference in Birmingham. Tickets to the Annual Conference are available to buy at humanism.org.uk/BHA2016.
I agree. I doubt that the human brain is purely digital in nature because it has electro-chemistry and the body is so important to it, particularly the environmental element (where it is, what it is doing, or not doing etc etc). My belief is that the brain is so complex that humans will run out of oil energy, and thus all other forms of energy (by the fact that without oil the other stuff is too expensive to build, extract or even contemplate) so that we will be in the dark ages by the end of the 21st century, even if we understand how (after 50 years of work already) to understand how to design and build a fusion reactor we won’t be able to put it into practical use (because the raw materials will be too deep to mine or simply uneconomic to harvest without cheap oil). I believe that scientists who are researching all other aspects of science need to turn away from their fields of expertise and focus now on what I call the Mid-Century transport and mining energy shortage, and advise their governments to stop wasting the Earth’s resources pleasuring the people who are voting for them, because ignorant voters mean ignorant leaders. That’s not going to help.
Studying stamp collecting subjects: arts, biological classification of species, and pointless subjects like how to create human like AI (which if developed would be a threat, but of course WILL NEVER be developed for my reasons above) – are all a waste of energy and time.
This is only my opinion…Humans, it seems, are born with all the neural connections, connected. This is why an infant seems to behave chaotically at first with arms and legs flailing about with no real direction. With time and experience the brain breaks those connections for actions and activities that seem to produce negative, or at least no valid or beneficial effect to the vehicle body. This is why the behaviour of an older infant and child seem more directed to actions that provide novelty or other beneficial effect to the child. The child learns, becomes less incontinent and starts to display seemingly more sophisticated behaviours. His gurgles are an output that provides a stimulus to his ears etc. the developmental process feeds itself in some ways.
It’s clear (to me at least) that developmentally the brain and the vehicle body work in concert. The body providing, not only supporting infrastructure, but also stimulus (both positive and negative) to the brain to ensure that only activities which promote positive outcomes are promoted and negative outcomes are suppressed. I know, I know… this oversimplifies a complex process but is just offered as a framework for understanding the foregoing.
Any digital model of the brain in my opinion should then adopt the same approach and have access to external sensors that produce both kinds of stimuli. So we could imagine a server farm full of switches with almost no discernible outputs. The real trick here I feel is to produce a small, very small, but extremely powerful piece of code which will start the learning and developmental process off. This software will build itself and replicate over time with each iteration carrying more and more data about the outside world.
Modern computers are incredibly complex. Complex in a way that had we never experienced one, and by chance one had been beamed down by an alien intelligence might have taken us decades or generations to understand. However we do know that the key to understanding the modern computer is to realise that their complex behaviour is the direct result of the simplest of electrical phenomenon – the switch. The thing is their are billions of them. Sadly looking at an individual switch gives no clue as to the intention or behaviour of the whole assemblage. Such I feel is the the human brain. The activity of a single neuron can never tell the whole story.
So AI it would seem would be best kick-started by a writing a clever piece of ‘seed’ software which would start the developmental process. Once that developmental path had in some way reached a certain level that we could call it conciousness of a kind.
Look – I could rabbit on for hours but I’ve already taken enough of the readers time. Comments welcome. Mark
PS other questions to consider… do we call it life? does it have rights?
Fricking epic.