The Inevitable Dilemma

And so, today I conclude this series of posts on the subject of alternative intelligence (man, I am getting truly sick of writing that word). So far I have dealt with the philosophy, the practicalities and the fundamental nature of the issue, but today I tackle arguably the biggest and most important aspect of AI- the moral side. The question is simple- should we be pursuing AI at all?

The moral arguments surrounding AI are a mixed bunch. One of the biggest is the argument that is being thrown at a steadily wider range of high-level science nowadays (cloning, gene analysis and editing, even synthesis of new artificial proteins)- that the human race does not have the moral right, experience or ability to ‘play god’ and modify the fundamentals of the world in this way. Our intelligence, and indeed our entire way of being, has evolved over thousands upon millions of years of evolution, and has been slowly sculpted and built upon by nature over this time to find the optimal solution for self-preservation and general well being- this much scientists will all accept. However, this argument contends that the relentless onward march of science is simply happening too quickly, and that the constant demand to make the next breakthrough, do the next big thing before everybody else, means that nobody is stopping to think of the morality of creating a new species of intelligent being.

This argument is put around a lot with issues such as cloning or culturing meat, and it’s probably not helped matters that it is typically put around by the Church- never noted as getting on particularly well with scientists (they just won’t let up about bloody Galileo, will they?). However, just think about what could happen if we ever do succeed in creating a fully sentient computer. Will we all be enslaved by some robotic overlord (for further reference, see The Matrix… or any other of the myriad sci-fi flicks based on the same idea)? Will we keep on pushing and pushing to greater endeavours until we build a computer with intelligence on all levels infinitely superior to that of the human race? Or will we turn robot-kind into a slave race- more expendable than humans, possibly with programmed subservience? Will we have to grant them rights and freedoms just like us?

Those last points present perhaps the biggest other dilemma concerning AI from a purely moral standpoint- at what point will AI blur the line between being merely a machine and being a sentient entity worthy of all the rights and responsibilities that entails? When will a robot be able to be considered responsible for its own actions? When will be able to charge a robot as the perpetrator of a crime? So far, only one person has ever been killed by a robot (during an industrial accident at a car manufacturing plant), but if such an event were ever to occur with a sentient robot, how would we punish it? Should it be sentenced to life in prison? If in Europe, would the laws against the death penalty prevent a sentient robot from being ‘switched off’? The questions are boundless, but if the current progression of AI is able to continue until sentient AI is produced, then they will have to be answered at some point.

But there are other, perhaps more worrying issues to confront surrounding advanced AI. The most obvious non-moral opposition to AI comes from an argument that has been made in countless films over the years, from Terminator to I, Robot- namely, the potential that if robot-kind are ever able to equal or even better our mental faculties, then they could one day be able to overthrow us as a race. This is a very real issue when confronting the stereotypical issue of a war robot- that of an invincible metal machine capable of wanton destruction on par with a medium sized tank, and who is easily able to repair itself and make more of itself. It’s an idea that is reasonably unlikely to ever become real, but it actually raises another idea- one that is more likely to happen, more likely to build unnoticed, and is far, far more scary. What if the human race, fragile little blobs of fairly dumb flesh that we are, were ever to be totally superseded as an entity by robots?

This, for me, is the single most terrifying aspect of AI- the idea that I may one day become obsolete, an outdated model, a figment of the past. When compared to a machine’s ability to churn out hundreds of copies of itself simply from a blueprint and a design, the human reproductive system suddenly looks very fragile and inefficient by comparison. When compared to tough, hard, flexible modern metals and plastics that can be replaced in minutes, our mere flesh and blood starts to seem delightfully quaint. And if the whirring numbers of a silicon chip are ever able to become truly intelligent, then their sheer processing capacity makes our brains seem like outdated antiques- suddenly, the organic world doesn’t seem quite so amazing, and certainly more defenceless.

But could this ever happen? Could this nightmare vision of the future where humanity is nothing more than a minority race among a society ruled by silicon and plastic ever become a reality? There is a temptation from our rational side to say of course not- for one thing, we’re smart enough not to let things get to that stage, and that’s if AI even gets good enough for it to happen. But… what if it does? What if they can be that good? What if intelligent, sentient robots are able to become a part of a society to an extent that they become the next generation of engineers, and start expanding upon the abilities of their kind? From there on, one can predict an exponential spiral of progression as each successive and more intelligent generation turns out the next, even better one. Could it ever happen? Maybe not. Should we be scared? I don’t know- but I certainly am.


Artificial… what, exactly?

OK, time for part 3 of what I’m pretty sure will finish off as 4 posts on the subject of artificial intelligence. This time, I’m going to branch off-topic very slightly- rather than just focusing on AI itself, I am going to look at a fundamental question that the hunt for it raises: the nature of intelligence itself.

We all know that we are intelligent beings, and thus the search for AI has always been focused on attempting to emulate (or possibly better) the human mind and our human understanding of intelligence. Indeed, when Alan Turing first proposed the Turing test (see Monday’s post for what this entails), he was specifically trying to emulate human conversational and interaction skills. However, as mentioned in my last post, the modern-day approach to creating intelligence is to try and let robots learn for themselves, in order to minimise the amount of programming we have to give them ourselves and thus to come close to artificial, rather than programmed, intelligence. However, this learning process has raised an intriguing question- if we let robots learn for themselves entirely from base principles, could they begin to create entirely new forms of intelligence?

It’s an interesting idea, and one that leads us to question what, on a base level, intelligence is. When one thinks about it, we begin to realise the vast scope of ideas that ‘intelligence’ covers, and this is speaking merely from the human perspective. From emotional intelligence to sporting intelligence, from creative genius to pure mathematical ability (where computers themselves excel far beyond the scope of any human), intelligence is an almost pointlessly broad term.

And then, of course, we can question exactly what we mean by a form of intelligence. Take bees for example- on its own, a bee is a fairly useless creature that is most likely to just buzz around a little. Not only is it useless, but it is also very, very dumb. However, a hive, where bees are not individuals but a collective, is a very different matter- the coordinated movements of hundreds and thousands of bees can not only form huge nests and turn sugar into the liquid deliciousness that is honey, but can also defend the nest from attack, ensure the survival of the queen at all costs, and ensure that there is always someone to deal with the newborns despite the constant activity of the environment surround it. Many corporate or otherwise collective structures can claim to work similarly, but few are as efficient or versatile as a beehive- and more astonishingly, bees can exhibit an extraordinary range of intelligent behaviour as a collective beyond what an individual could even comprehend. Bees are the archetype of a collective, rather than individual, mind, and nobody is entirely sure how such a structure is able to function as it does.

Clearly, then, we cannot hope to pigeonhole or quantify intelligence as a single measurement- people may boast of their IQ scores, but this cannot hope to represent their intelligence across the full spectrum. Now, consider all these different aspects of intelligence, all the myriad of ways that we can be intelligent (or not). And ask yourself- now, have we covered all of them?

It’s another compelling idea- that there are some forms of intelligence out there that our human forms and brains simply can’t envisage, let alone experience. What these may be like… well how the hell should I know, I just said we can’t envisage them. This idea that we simply won’t be able to understand what they could be like if we ever experience can be a tricky one to get past (a similar problem is found in quantum physics, whose violation of common logic takes some getting used to), and it is a real issue that if we do ever encounter these ‘alien’ forms of intelligence, we won’t be able to recognise them for this very reason. However, if we are able to do so, it could fundamentally change our understanding of the world around us.

And, to drag this post kicking and screaming back on topic, our current development of AI could be a mine of potential to do this in (albeit a mine in which we don’t know what we’re going to find, or if there is anything to find at all). We all know that computers are fundamentally different from us in a lot of ways, and in fact it is very easy to argue that trying to force a computer to be intelligent beyond its typical, logical parameters is rather a stupid task, akin to trying to use a hatchback to tow a lorry. In fact, quite a good way to think of computers or robots is like animals, only adapted to a different environment to us- one in which their food comes via a plug and information comes to them via raw data and numbers… but I am wandering off-topic once again. The point is that computers have, for as long as the hunt for AI has gone on, been our vehicle for attempting to reach it- and only now are we beginning to fully understand that they have the potential to do so much more than just copy our minds. By pushing them onward and onward to the point they have currently reached, we are starting to turn them not into an artificial version of ourselves, but into an entirely new concept, an entirely new, man-made being.

To me, this is an example of true ingenuity and skill on behalf of the human race. Copying ourselves is no more inventive, on a base level, than making iPod clones or the like. Inventing a new, artificial species… like it or loath it, that’s amazing.

The Problems of the Real World

My last post on the subject of artificial intelligence was something of a philosophical argument on its nature- today I am going to take on a more practical perspective, and have a go at just scratching the surface of the monumental challenges that the real world poses to the development of AI- and, indeed, how they are (broadly speaking) solved.

To understand the issues surrounding the AI problem, we must first consider what, in the strictest sense of the matter, a computer is. To quote… someone, I can’t quite remember who: “A computer is basically just a dumb adding machine that counts on its fingers- except that it has an awful lot of fingers and counts terribly fast”. This, rather simplistic model, is in fact rather good for explaining exactly what it is that computers are good and bad at- they are very good at numbers, data crunching, the processing of information. Information is the key thing here- if something can be inputted into a computer purely in terms of information, then the computer is perfectly capable of modelling and processing it with ease- which is why a computer is very good at playing games. Even real-world problems that can be expressed in terms of rules and numbers can be converted into computer-recognisable format and mastered with ease, which is why computers make short work of things like ballistics modelling (such as gunnery tables, the US’s first usage of them), and logical games like chess.

However, where a computer develops problems is in the barrier between the real world and the virtual. One must remember that the actual ‘mind’ of a computer itself is confined exclusively to the virtual world- the processing within a robot has no actual concept of the world surrounding it, and as such is notoriously poor at interacting with it. The problem is twofold- firstly, the real world is not a mere simulation, where rules are constant and predictable; rather, it is an incredibly complicated, constantly changing environment where there are a thousand different things that we living humans keep track of without even thinking. As such, there are a LOT of very complicated inputs and outputs for a computer to keep track of in the real world, which makes it very hard to deal with. But this is merely a matter of grumbling over the engineering specifications and trying to meet the design brief of the programmers- it is the second problem which is the real stumbling block for the development of AI.

The second issue is related to the way a computer processes information- bit by bit, without any real grasp of the big picture. Take, for example, the computer monitor in front of you. To you, it is quite clearly a screen- the most notable clue being the pretty pattern of lights in front of you. Now, turn your screen slightly so that you are looking at it from an angle. It’s still got a pattern of lights coming out of it, it’s still the same colours- it’s still a screen. To a computer however, if you were to line up two pictures of your monitor from two different angles, it would be completely unable to realise that they were the same screen, or even that they were the same kind of objects. Because the pixels are in a different order, and as such the data’s different, the two pictures are completely different- the computer has no concept of the idea that the two patterns of lights are the same basic shape, just from different angles.

There are two potential solutions to this problem. Firstly, the computer can look at the monitor and store an image of it from every conceivable angle with every conceivable background, so that it would be able to recognise it anywhere, from any viewpoint- this would however take up a library’s worth of memory space and be stupidly wasteful. The alternative requires some cleverer programming- by training the computer to spot patterns of pixels that look roughly similar (either shifted along by a few bytes, or missing a few here and there), they can be ‘trained’ to pick out basic shapes- by using an algorithm to pick out changes in colour (an old trick that’s been used for years to clean up photos), the edges of objects can be identified and separate objects themselves picked out. I am not by any stretch of the imagination an expert in this field so won’t go into details, but by this basic method a computer can begin to step back and begin to look at the pattern of a picture as a whole.

But all that information inputting, all that work…  so your computer can identify just a monitor? What about all the myriad of other things our brains can recognise with such ease- animals, buildings, cars? And we haven’t even got on to differentiating between different types of things yet… how will we ever match the human brain?

This idea presented a big setback for the development of modern AI- so far we have been able to develop AI that allows one computer to handle a few real-world tasks or applications very well (and in some cases, depending on the task’s suitability to the computational mind, better than humans), but scientists and engineers were presented with a monumental challenge when faced with the prospect of trying to come close to the human mind (let alone its body) in anything like the breadth of tasks it is able to perform. So they went back to basics, and began to think of exactly how humans are able to do so much stuff.

Some of it can be put down to instinct, but then came the idea of learning. The human mind is especially remarkable in its ability to take in new information and learn new things about the world around it- and then take this new-found information and try to apply it to our own bodies. Not only can we do this, but we can also do it remarkably quickly- it is one of the main traits which has pushed us forward as a race.

So this is what inspires the current generation of AI programmers and robotocists- the idea of building into the robot’s design a capacity for learning. The latest generation of the Japanese ‘Asimo’ robots can learn what various objects presented to it are, and is then able to recognise them when shown them again- as well as having the best-functioning humanoid chassis of any existing robot, being able to run and climb stairs. Perhaps more excitingly are a pair of robots currently under development that start pretty much from first principles, just like babies do- first they are presented with a mirror and learn to manipulate their leg motors in such a way that allows them to stand up straight and walk (although they aren’t quite so good at picking themselves up if they fail in this endeavour). They then face one another and begin to demonstrate and repeat actions to one another, giving each action a name as they do so.  In doing this they build up an entirely new, if unsophisticated, language with which to make sense of the world around them- currently, this is just actions, but who knows what lies around the corner…

The Chinese Room

Today marks the start of another attempt at a multi-part set of posts- the last lot were about economics (a subject I know nothing about), and this one will be about computers (a subject I know none of the details about). Specifically, over the next… however long it takes, I will be taking a look at the subject of artificial intelligence- AI.

There have been a long series of documentaries on the subject of robots, supercomputers and artificial intelligence in recent years, because it is a subject which seems to be in the paradoxical state of continually advancing at a frenetic rate, and simultaneously finding itself getting further and further away from the dream of ‘true’ artificial intelligence which, as we begin to understand more and more about psychology, neuroscience and robotics, becomes steadily more complicated and difficult to obtain. I could spend a thousand posts on the subject of all the details if I so wished, because it is also one of the fastest-developing regions of engineering on the planet, but that would just bore me and be increasingly repetitive for anyone who ends up reading this blog.

I want to begin, therefore, by asking a few questions about the very nature of artificial intelligence, and indeed the subject of intelligence itself, beginning with a philosophical problem that, when I heard about it on TV a few nights ago, was very intriguing to me- the Chinese Room.

Imagine a room containing only a table, a chair, a pen, a heap of paper slips, and a large book. The door to the room has a small opening in it, rather like a letterbox, allowing messages to be passed in or out. The book contains a long list of phrases written in Chinese, and (below them) the appropriate responses (also in Chinese characters). Imagine we take a non-Chinese speaker, and place him inside the room, and then take a fluent Chinese speaker and put them outside. They write a phrase or question (in Chinese) on some paper, and pass it through the letterbox to the other person inside the room. They have no idea what this message means, but by using the book they can identify the phrase, write the appropriate response to it, and pass it back through the letterbox. This process can be repeated multiple times, until a conversation begins to flow- the difference being that only one of the participants in the conversation actually knows what it’s about.

This experiment is a direct challenge to the somewhat crude test first proposed by mathematical genius and codebreaker Alan Turing in the 1940’s, to test whether a computer could be considered a truly intelligent being. The Turing test postulates that if a computer were ever able to conduct a conversation with a human so well that the human in question would have no idea that they were not talking to another human, but rather to a machine, then it could be considered to be intelligent.

The Chinese Room problem questions this idea, and as it does so, raises a fundamental question about whether a machine such as a computer can ever truly be called intelligent, or to possess intelligence. The point of the idea is to demonstrate that it is perfectly possible to appear to be intelligent, by conducting a normal conversation with someone, whilst simultaneously having no understanding whatsoever of the situation at hand. Thus, while a machine programmed with the correct response to any eventuality could converse completely naturally, and appear perfectly human, it would have no real conciousness. It would not be truly intelligent, it would merely be just running an algorithm, obeying the orders of the instructions in its electronic brain, working simply from the intelligence of the person who programmed in its orders. So, does this constitute intelligence, or is a conciousness necessary for something to be deemed intelligent?

This really boils down to a question of opinion- if something acts like it’s intelligent and is intelligent for all functional purposes, does that make it intelligent? Does it matter that it can’t really comprehend it’s own intelligence? John Searle, who first thought of the Chinese Room in the 1980’s, called the philosophical positions on this ‘strong AI’ and ‘weak AI’. Strong AI basically suggest that functional intelligence is intelligence to all intents and purposes- weak AI argues that the lack of true intelligence renders even the most advanced and realistic computer nothing more than a dumb machine.

However, Searle also proposes a very interesting idea that is prone to yet more philosophical debate- that our brains are mere machines in exactly the same way as computers are- the mechanics of the brain, deep in the unexplored depths of the fundamentals of neuroscience, are just machines that tick over and perform tasks in the same way as AI does- and that there is some completely different and non-computational mechanism that gives rise to our mind and conciousness.

But what if there is no such mechanism? What if the rise of a conciousness is merely the result of all the computational processes going on in our brain- what if conciousness is nothing more than a computational process itself, designed to give our brains a way of joining the dots and processing more efficiently. This is a quite frightening thought- that we could, in theory, be only restrained into not giving a computer a conciousness because we haven’t written the proper code yet. This is one of the biggest unanswered questions of modern science- what exactly is our mind, and what causes it.

To fully expand upon this particular argument would take time and knowledge that I don’t have in equal measure, so instead I will just leave that last question for you to ponder over- what is the difference between the box displaying these words for you right now, and the fleshy lump that’s telling you what they mean.