Artificial intelligence singularity, ‘singularity’ presumably borrowed from cosmology, is the inevitable point at which artificial intelligence (AI) becomes equivalent to, and then goes on to surpass, human intelligence (HI). Most scientists and technology experts seem to have little doubt that this will happen soon, most likely in the next 50 to 100 years or so. They’re wrong.

They’re wrong because of the way they appear to define ‘human intelligence’, and because that definition is going to be critical in any discussion around the idea of AI singularity, it has to be an accurate, useful and fair one.

It seems scientists think that human intelligence is only manifest in our ability to think logically; our ability to rationalise, our ability to reason. But although we are extremely good at those things, that is a very narrow and naive way to view human intelligence. We could take this narrow view a stage further and say that AI has already surpassed HI, if we viewed that to be the ability to crunch numbers, or the capacity to hold and recall a mass of newly acquired data in memory. Is an AI’s ability to beat us at chess, or multiply two random six-digit numbers in a micro-second, in any way significant? No, because that’s not the primary purpose of our ability to form ‘memories’. Accurate recall is rarely what’s required.

We Should Compare Like-With-Like

Human intelligence isn’t about our ability to understand that 1+1=2. That’s not what we’re designed to do. It’s helpful, but it’s not really that important.

Our intelligence is born from an ancient evolutionary need; essentially as a ‘sexed’ species, we need to attract the best possible mate to make and raise children with, to protect and nurture those children to adulthood, and avoid death at least until those tasks are completed. And as anyone can see, we’re extremely good at that – some would say too good.

It is these fundamental needs that has led humans to master an enormous range of skills, abilities and knowledge that go way beyond just logic and reason. So in order to claim AI singularity, it must be shown that AI has surpassed humans in all these other areas too. After all, if we’re not going to compare like-with-like, then why compare at all?

Emotion (Motion, Motivation)

AI then, like us, will require some kind of ‘need’, something to give it the drive, the desire, for gaining experiences and knowledge. Experiences and knowledge that it ‘feels’ are ‘valuable’.

Gathering knowledge and learning is an innate desire in us, it’s a default state. Obtaining knowledge and experiences ‘mean’ something to us and we’re rewarded when we find them. We gain ‘satisfaction’ and ‘pleasure’ in doing all these things. We have emotions.

Our highly developed emotions are very much part of our unique intelligence, and we couldn’t function at all without them. Every thing we do, and every decision we make is reliant on them.

Every task that you attempt to tackle today will be affected by how you feel at that moment; your emotional state. Our emotions can’t be turned off.

We pursue that which gives us satisfaction, whilst trying to avoid that which does not. These pursuits determine to a large degree, the path our lives take. Our emotions are far more important, and have a far greater influence on our lives, than logic or reason do.

AI, without doubt, is going to need something similar. AI with desires that can be satisfied? How?

We might spend a couple of hard hours climbing a hill, just to ‘experience’ the view. As we approach the summit, excitement will rise in anticipation of what we might see. At the top we’ll likely smile in wonder at the beauty of that vista. An excited artificial intelligence, ‘willing’ to climb that hill?

Bodily Functions

We not only have the desire to learn, to gather knowledge and gain experiences, we are also very well equipped when it comes to going out and getting them. The human body is flexible, strong, dexterous, efficient, self-repairing and adaptable. Our truly wondrous physical design allows us to traverse practically any terrain, and we’ve gone on to explore virtually every corner of planet Earth and beyond. Our physical adaptability also affords us an ability to train for multiple, diverse, high-skilled activities – think of the gymnast, the concert pianist, the brain surgeon.

Many theories suggest our highly evolved intellectual capabilities were brought about in tandem with our physical evolution – for example, us walking upright or our front-facing eyes with advanced colour and depth perception.

There can be little doubt there’s one part of our anatomy that must have evolved in sync alongside our brains – our hands. Our hands can manoeuvre and be manipulated to perform endless tasks, and the fine control our brains develop in them is beyond compare – there’s no other control system that even comes remotely close.

Our hands are our ‘feelers’, they feedback information that help us make sense of the world. Perhaps just as importantly though, they allow us to turn our thoughts into practical reality. Our ability to make things allows us to consider the success of a concept or how it might be improved. The practical and prolonged use of our made objects teach us things that could never be understood by any other means. Materials, stresses, structures, strengths, efficiencies, etc. Whole disciplines have developed from us making things – engineering, architecture, chemistry. We even have historical periods named after the things we make – the Iron Age and the Bronze Age.

Our hands teach us, they bring about new synaptic connections in our brains, they make us more intelligent.

So any AI system will need to be mobile, with built-in tools allowing it to explore, understand and put ideas into action. Looking at current technologies, this will presumably be some kind of robotic design. A robotic design that can surpass the bone, cartilage, muscle, and the nerve control system that we have?

Have Some Sense

On your face you have two very clever, yet simple, radiation sensors. We’ve been making artificial versions of these sensors for over a 100 years. Our sensors work brilliantly for us, yet when we attempt to combine the artificial ones with AI – absolute chaos ensues. Really laughable when you see it, and a great example of scientific presumptuous arrogance.

The way our eyes and brains work together to interpret ‘reality’ is extremely complex and mysterious. Our eyes receive the information (light), but it’s our brains that see. Visual recognition and interpretation is where the complexity lies. Our brains perform so many operations on the process of ‘seeing’, that our other senses are sometimes held back to preserve perceived synchronicity.

The visual information coming to our brains isn’t really smooth and continuous, it’s more akin to a series of photographic stills. But our brains fill in the gaps. Much of what we ‘see’ is just our brain making assumptions about what it is in front of us. Human faces protrude toward us, so naturally our brains refuse to ‘recognise’ one that doesn’t.

How do you recognise a house? Is it size, colour, shape, architectural style, building materials, usage, position, age? That it has windows, the size and number of windows, what’s in the windows? Is it still a house if no one lives in it, what if the roof is also missing? A small model of a house made from Lego? A drawing?

Thousands of scientific papers have been written on human visual perception, with layer on layer of complexity being newly discovered. Yet still, little is really understood.

Emulating just this single human sense will be a massive task, and that’s ignoring all the other sensory abilities we humans enjoy.

The deep complexity of the human multi-sensory interface, and the way all that information is processed by our brains, is currently way beyond our understanding. So how likely is it we will be able to manufacture a sensory system that’s similarly capable in the next hundred years?

“Sorry, I don’t understand the question?”

One of many stock answers, real or fictional, given by AI’s to indicate a fundamental communication failure.

We possess the means to vibrate air in thousands of different ways, our ability to control these vibrations means we can repeat the same ‘sounds’ again and again. And because our brains are so wondrously intelligent we are able to attach meaning to these noises we make. We can ‘speak’, we have language, we can communicate.

Just by throwing noises at each other we are able to transfer complex ideas, such as “why a cat can be both dead and alive at the same time”, from one brain to another.

We take this remarkable ability for granted, because we can do it with such ease. But like so much else that we do, speech and language recognition and comprehension is far from simple. Many notable intellects, such as Wittgenstein, Chomsky and Pinker have all offered various explanations as to how it might work, but none are really seen as definitive. An intellectual ability that we don’t yet understand in ourselves, let alone developing an artificial equivalent.

With speech, it’s not just the words that carry meaning either. It’s why artificial speech sounds so strange to us, because when we speak we add expression, phrasing and pitch, which when combined can convey a different meaning to what we say.

Spoken language is just one of many means by which we communicate. A close relative of course is written language. Most often written languages are just a series of shapes that translate to the sounds we speak, phonetic like English, but sometimes written and spoken languages can be unrelated. It shouldn’t be forgotten though, that reading and writing and quite different from speaking and hearing – different production, different senses. Further complex processing our brilliant brains have to cope with.

As emotional beings, it’s important in our social interactions that we have some means to gauge how others around us are feeling. This is where ‘body language’ comes in to play, our facial expressions in particular. Just by looking at muscle tensions our brains can ‘see’ anger, happiness, sadness, fear, surprise, confusion, disgust, wonder, coyness, despair, relief, love and hate… in a face.

Lastly there’s all the abstract, subtle and allegorical ways our brains communicate. Something is communicated through the clothes we wear, or the way we decorate our homes. We communicate via the arts – poetry, film, sculpture and music. So much of what we do is an attempt to say something about ourselves; what we represent, what we feel, signalling to others who we are.

So advanced are our communicative abilities, we can even use them to deceive and manipulate. No easy feat.

The elaborate, highly sophisticated, multi-layered ways we humans communicate with each other, is so far beyond anything we can currently even imagine replicating artificially.

And Lastly… We Die

You know you’re going to die, but does a fish? Or a goat, or a dog? Perhaps a bonobo? Probably not.

So here we are at last, with one of the big impenetrable questions – consciousness.

Most experts agree we have consciousness, but there’s less agreement about what it is and how it works. Some suggest any kind of intelligence, if it can continue to develop, will at some stage become self-aware. It’s a kind of ‘crowning glory’, the ‘sum of all parts’, an inevitable point that we, as highly intelligent beings, so happen to have reached. Others feel that our consciousness may not be a physical thing at all, and therefore can’t really reside in brains, or anything else. The ‘brain’ and ‘mind’ conundrum.

It is a tricky area, because consciousness is a kind of special case in ‘thought’. But what is a thought? Where does it reside? What happens to a thought when we stop thinking it? We as conscious creatures, have the ability to think about what we’re thinking, and then think about the fact that we’re thinking about what we’re thinking… etc.

Consciousness is obviously an important aspect of our intelligence, and as far as we know, it’s a unique possession. Our ability to self-reflect probably had a large part to play in the world’s religions and important areas of philosophical thought. Similarly with ideologies, ‘human rights’ and socio-political systems. Consciousness may be the agency necessary to bring about all these things.

Certainly, no AI system that lacks consciousness, can ever be considered our intellectual equal.

Belittles Our Brains

Comparisons between AI and HI have to be made on a level playing field. Showing that AI can do things better in areas that we’re not ‘designed’ to excel in – areas that aren’t important to us – is pointless and unfair.

Scientists and technologists want to shoe-horn us into their model of intelligence – that we’re some kind of machine-like, computer-like entity. But we’re not like that at all.

Even the parts that seem similar – logic, rationality – may not be. We’ve as yet no way of telling. Some suggest that our ‘rational’ development didn’t even come about as a means to explore objective truth, but was instead just a clever ‘trick’ for us to get our own way. Brain verses brawn.

Being narrowly selective in defining intelligence like this, has no real comparative value. But it does pose a danger, in that there’s an underlying message that suggests ‘rational thought’ is the only ‘quality’ worth measuring against. We’ve been surpassed because all the other non-rational abilities we have are irrelevant, extraneous and unimportant.

Think of the important things that have happened in you life, the things your memory was designed to remember? Or think of the things that matter to you most and why? What part does rationality play?

True AI singularity – not in a million years.