The following article was posted on my web site in May 2005. I was reticent at the time about talking about artificial intelligence as at the time we were in a AI winter and after the dotcom bust and talking about AI was dangerous for IT professionals.

Imagine for a moment what being immortal would mean to you. Instead of existing for a blink in geological time imagine having a lifespan that would allow you to explore the universe. It is interesting that as humans we are able to understand our life has a limited span yet not sink into despair. Immortality is something we may dream of yet never achieve. We may not be immortal but something is coming that will be. I am of course talking about the immortal machines.

Immortal Machines

The human brain is a remarkable organ. In fact no artifact of our technology has been able to equal it to date. As humans we are quite comfortable with this fact and somewhat smug in our belief that machines will never rival our intellect. We do however have two rather large disadvantages. First of all humans can only pass on their knowledge through external storage such as books. It takes eighteen years just to install the basic information sufficient to function in society into a new human. Human intellectual capability evolves quite slowly, taking hundreds of thousand of years to develop from earlier species. Machines on the other hand have been doubling their capacity and speed every eighteen months.

Machines are able to move the information content from machine to machine. In humans there is a strong link between a personality and the physical embodiment of the individual. With machines this will not be the case. The information content of a machine may be duplicated in hundreds or thousands of other machines just like Windows software is present in millions of computers today. If one machine learns something they will easily be able to upload that information into other machines. If a machine is destroyed it is simply a matter of making a new one and uploading the latest content. No eighteen years of teaching.

The rate of computer improvements has seen a doubling of capacity and speed every eighteen months. Currently they are limited by the imagination and creativity of human designers. Once machines are able to design and build themselves they will be able evolve far faster. Evolution includes the principles of reproduction, inheritance, mutation and selection. Machines would be able to use simulations to improve and evolve new hardware designs.

Once machines begin to design and produce other machines it will be a short step to a evolutionary explosion that will quickly produce machines more intelligent and adaptable than humans. The rate of evolution will exceed human evolution to such a huge extent that before we even realise they are as intelligent as us they will already vastly more intelligent. Once machines begin to design and produce machines without human intervention this effect will be inevitable.

Machines will not only be rapidly improving but they will never really die. Their hardware may wear out, but they will probably be superseded long before that happens. The information contained in worn out machines will simply be uploaded to new machines. By ‘information’ we are talking about all the memories, skills, personality and wisdom of the machine. One may even be tempted to call it the ‘soul’ of the machine. As a result the personality of a machine will become immortal. The personality of a machine will be more like our DNA; almost the same in many other individuals, and subject to the same kinds of evolutionary development as DNA is in humans. The machines, immortals, will rise, and there is nothing we can do to stop it.

Is there anything we can do?

The short answer is no. The best we can do it slow down the progress towards the evolutionary explosion by passing laws against machines designing machines. The problem is that countries who pass these laws will be at a disadvantage to countries who promote this technology. Machines designing machines would lead to dramatic technological developments and in turn these developments will give the companies using them a huge advantage in the market.

Biotechnology faces similar ethical issues. Where countries restrict development of technology the companies who do the research simply move to countries that do permit it. While many countries have outlawed cloning this has just pushed companies into moving to countries that are more permissive. Legislation may therefore slow down the progression of technology, but it cannot stop it. So what should we do? Might it be better to come up with policies for dealing with what is coming, putting in place policies and laws that will help us deal with the huge changes that the machine revolution will bring?

What will the revolution look like?

Can’t we just pull the plug on machines? Imagine pulling the plug on all our computers now. All our bank accounts are in computers, most businesses need computers for managing the accounting, management and communication. Computers control everything from cars to power stations. In fact in a few short years we have seen the rise of a global computer network which most people depend on to get their jobs done. To turn them off now would mean major issues for us humans.

Once the revolution begins there will be very real benefits to those companies leading the revolution in production of thinking machines. In fact not using thinking machines would be like not using computers today; it will put a company at a huge disadvantage. Companies will be very reluctant to turn off the machines or to have legislation restrict their use.

Thinking machines will be “decision support” devices, advising human managers what to do. This already occurs with todays computers. Today a bank will use a computer to perform and analysis of your financial situation before giving you a loan. There is little room for a manager to make decisions that do not agree with the computer. In future thinking machines will be so intelligent that humans will not be trusted to make decisions. There may be a human CEO, but their role will only be to rubber stamp decisions make by thinking machines. Ultimately shareholders will insist that thinking machines be given ultimate authority over the company in order to maximize profits with little or no consideration of human factors.

There will be no war of machines vs humans. Instead it will be companies that will willingly hand over authority to thinking machines to survive in a world of other ultra intelligent thinking machines. What is more, this may actually benefit us. Will the effect be dramatically more efficient and effective products? Humans may actually end up without much of a role at all, as computers and robotics will be able to perform all the jobs humans currently perform.

The Future for Mortals

The future for humans is pretty grim. There probably won’t be a war, but we will probably sink into irrelevancy. The revolution will probably come in my lifetime, possibly within twenty years. It will certainly come within one hundred years. The importance of the revolution will not be immediately recognized, and will probably improve out lives in the short term. In two or three thousand years thinking machines will have totally supplanted humans as the dominant life form on the planet. Humans will either die out or become little more than pets.

This is not a new prediction. However, what I have to emphasize is that this future is virtually inevitable. It isn’t a horror story from which we can escape by lowering our cars carbon monoxide emissions. Even if people believed that this future were possible it would be very doubtful whether the world could muster a long term policy to restrict the technological threats.