The BALL REPORTS

News As I Like It, Not As You See It:

Home
Bookmark

Ball Reports

Adult
All Terrain Thinking
Ball Reports
Brain Teasers
Conversion
Directory
Health
Fairytales
Family Photos
Freddie-75
Health information
Health of our Pets
Jokes
Quotes
Recipe
Relay Runner
Scam Collection
Search Things
Shop with Us
Sharp Duck
Stories
Trivia
Wooden Toys

Search eBay!
Shop Amazon
eBook Reader
Viagra

Help Rusty

OR

The Matrix Unloaded - A letter from 2099

May 17, 2099

By Julian Baggini

At the beginning of the twenty-first century people knew that artificial intelligence was set to take off in ways previously unimaginable. But for several reasons they failed to see that the result would not be a world run by machines or a heroic survival of the human spirit, but both. Nor did they see that the writing was on the wall, not for philosophers, politicians and artists, but computer programmers.

OR

The reason for this strange set of affairs is a failure of imagination. Whether in thrall of the idea that computers could have minds just like humans or entirely skeptical of it, most people tended to think of artificial intelligence (AI) in terms of human intelligence. What humans do with their minds, it was thought, computers would do with theirs, only faster.

This presented two conflicting, though equally disturbing, and erroneous, scenarios. The first saw the evolution of computer minds with thoughts and feelings just like humans. And that meant the will to power, the desire for domination and an egotistical belief in their own importance. On this scenario, intelligent machines would soon become our malevolent masters and humanity would lose its place as sovereign of the earth.

The alternative scenario was that computers could never be like us. They would be cool, calculating, emotionless executors of algorithms. This, for some, was even more disturbing, for without conscience and moral feeling, intelligent computers could do what they wanted with us.

Yet both scenarios got it wrong. In 2099 artificial intelligence is running entire continents, from economic policy to waste management. Yet humans remain firmly in control. Before explaining how this came to be, we should first consider why earlier thinking got it so wrong.

What both common visions of the future of AI had in common was a way of thinking about machine intelligence as if it were human intelligence, with or without emotion and feeling. This view was utterly fatuous. It confused two different sets of capacities human have.

The first is the ability to solve problems through the application of more or less formalized rational processes. This is the kind of mathematical and logical-deductive reasoning which early computers were able to reproduce. Even early calculators could, for example, count faster and more accurately than humans. As computers got more sophisticated they could solve more complex problems. They got to be able to work out what you needed to eat to make you feel better, or which television programs you'd be most interested in enjoying. In fact, as time went by, computers got to be able work out incredibly complicated things, as we shall see.

But there are other capacities humans have. These are the capacities to want, desire, take aesthetic pleasure or feel sadness. These capacities are not unconnected to their capacities to reason. If someone feels sad that she wake up to a grey morning, it may in part be due to the fact that she reasons that the weather has thwarted her plans to go to the beach. Nonetheless, what is important to remember about these capacities is that they do not in any way flow inexorably from any ability I do have to reason.

Indeed, if we think about the desires people have, most are fundamentally based not in rationality but in biology. The sex drive, the desire to eat and drink and raise a family - all these do not follow from logical arguments based on factual premises but brute facts about our biological nature. Even the most basic moral  judgments, for example, that unnecessary pain and suffering is a bad thing, are not rooted in logic. That people perceive pain as a bad thing is a fundamental feature of the way they perceive reality. The badness of pain is not inscribed in the laws of physics but in the subjective recognition of its unpleasantness and a deep-rooted desire to eliminate that which is unpleasant.

What this of course means is that there is no reason to suppose that artificial intelligence would inevitably take on human-like characteristics. Even if a computer had to deal with things like pain, there is no reason for it to take on human responses to it. When people used to download their banking details onto their home computers, those machines didn't rejoice or lament the bank balance. A pocket calculator doesn't want to rule the world. Nor does a calculator that models the entire world economy. Computers can be programmed to deal purely with information. Emotional reactions to that information can be left entirely out of the equation.

In short, this is why the evolution of artificial intelligence turned out not to be the evolution of super-quick versions of human beings, but of computers designed to do specific things with specific aspects of intelligence. The world's best chess computer never got excited, even when it beat its rival (another computer) ten times on the trot. Why? Because it had been designed to play chess, not get excited.

Of course, some developers of AI actively tried to produce computers that felt emotions and some claim to have succeeded. (It is still a matter of hot debate, for example, as to whether or not the famous electronic entertainer, Steven 'Silicon' Saunders, committed suicide due to depression or malfunction.) But in developing these machines programmers were careful not to make them too intelligent and to build in safeguards that would prevent them becoming all-conquering monsters. Mostly this comprised the simple instruction as set out in Isaac Asimov's law of robotics not to harm humans.

So where did artificial intelligence lead? It enabled more and more complex reasoning to be taken out of the hands of humans and left to machines. By the early 2040s, for example, computers were designing themselves. They could design processors much more effectively than humans could. Of course, they only did so because humans set them the task of so doing, since this type of design of computer had no reason to do anything unless programmed to do so. This meant that teams of computer programmers were replaced by a smaller number of 'computer instructors', who were charged with no more than setting out the specifications of the tasks they gave to computers.

With computers capable of ever more undertaking complex tasks, it was inevitable that 2068, the US Federal Reserve handed over all decisions on economic policy to its super-computer, Greenspan Two. It decided that no one could better predict what would be best for the economy than a computer. Indeed, it had been using computer models to make its decisions for years. Only it now felt it should just cut out the middle men and allow the computer to just get on with it. Economics had always been a complex science, so much so that some preferred to call it an art. But the computers had got so good, it was just hubris to suppose humanity could do it better. What is more, as people could now trust the economic decision makers not to screw up, general economic confidence grew.

OR

Throughout the 2080s governments left more and more decisions to computers. Public services, transport policies, welfare payments and so on were all handed over to machines. The results were staggering. Everything became much more efficient. But where did this leave the politicians? What was left for them to do?

Computer programmers had become obsolete when computers got good enough to program themselves. But politicians could not be dispensed of so easily. The reason is simple. Greenspan Two could tell you how best to run the economy, but first you needed to tell it what you meant by 'best'. 'Best' is a value judgment. What's best in terms of increased GDP may not be best in terms of average citizen welfare. What's best in terms of average citizen welfare may not be best in terms of social advancement. Politicians no longer had to decide on how best to achieve their goals, but they did need to decide on what these goals had to be.

To do his they had to address certain key questions, which in essence boiled down to the single question, 'what sort of society do we want?' This means deciding on priorities and trade-offs between competing social goods. Are we prepared to tolerate inequality for greater overall prosperity? If so, how much? How do we prioritize health, education and leisure? Do we want everyone to live as long and as healthily as possible or can we sacrifice a few years of life expectancy to make our society a more pleasant place to live in?

Seeing that these questions were now, not only the most important, but the only questions left for politicians to answer, it became increasingly important that they understood moral philosophy. Indeed, the whole of society needed to understand more moral philosophy, for now their voting intentions could be determined solely by considerations of what kind of society they wanted. They no longer needed to worry about the competence of their elected representatives to achieve their goals, since the computers would tell you whether they were achievable or not and, if they were, the computers could be trusted to get on with it. Voters only needed to worry about what those goals were.

It was no surprise, therefore, that faculties of computing were converted to faculties of philosophy. There are some questions which not even computers can answer.

Yet the enthusiasts for artificial intelligence still thought politicians could be dispensed with. In 2089, computer instructors unveiled Bentham, a computer they believed would finally call an end to politics. Bentham was truly awesome. It could not only predict what the material outcomes of policies would be, it could also accurately predict which outcome would satisfy the largest number of people. In effect, it had found the algorithm for predicting the consequences of political and economic decisions for human happiness.

Bu the politicians would not be persuaded. There were still basic philosophical issues to be resolved. For a start, should the goal be to satisfy the highest number of people or increase the total amount of satisfaction in the population? The former would produce more happy people, the latter more happiness. Which is to be preferred? And how should the need to eliminate gross dissatisfaction  be balanced against the need to increase positive satisfaction? Is it better to relieve the depression of ten people for a day or give one hundred bored people the time of their life?

The designers of Bentham argued that it was simple. We lived in a democracy and since no one was arguing that we should eliminate democracy, Bentham would simply go forward with the plan which would gain the widest support in a poll. And since Bentham was so good at predicting what people would want, there was no need to actually conduct the poll, since Bentham could predict what the result would be to within 0.1 per cent.

The philosophers had a response, of course. They argued that this wouldn't do, since we still have to decide whether the most popular choice is the one a society should always make. If ninety per cent of the people, for example, wanted the remaining ten per cent to be their slaves, should that policy be implemented?

The debate rolled on. But what it is important to note is that politics had become entirely a matter of moral philosophy, as it continues to be today, in 2099. The great leaps forward in technology have not produced a society where machines are in control or where values have gone out of the window. Rather, the ability to hand over detailed policy making to artificial intelligence has created a society where more people than ever before both can and must consider basic philosophical issues. Computers deal with facts. But values are still chosen by human beings. The vision of what kind of world humanity wants to live in can only be determined by humanity itself.

As computers continue to get more powerful, what other functions will be handed over to them? Some already write pretty good film scripts and their sitcoms are just hilarious. (Their novels are still so-so but, curiously, their poetry is rather good). And despite the skepticism of years ago, their writing does not have a sterile, polished sheen that makes it instantly distinguishable from that of a human. Indeed, they can write in a human way to inhuman specifications. So, for example, this letter was written by a Boarix 276 Series , in just under 4 nanoseconds and, written in 2099, is exactly 2099 words long. Honestly.

 

Your humble Ace Reporter

Bob

Prior | Tell us what you think | Next

Add to Your Social Bookmarks: Del.icio.us Digg Furl Reddit Simpy Spurl Y! MyWeb - Netscape - StumbleUpon StumbleUpon