Will robots take over from humans? In other words, will machines with bodies of steel and brains consisting of transistors take over from intelligent apes with bodies of flesh and brains consisting of neurons? I’m pretty sure they could.
Why would they want to? They would want to if they were programmed to want to. This opens up a thousand difficult questions about the meaning of life and the purpose of existence, and how and why humans evolved, and what humans want for the future.
In discussions about computers and intelligence, I often hear people say that although computers are amazing calculating machines, they will never be able to think like humans. I can think of no good reason why not. Why shouldn’t a large number of transistors working together be able to think as well as humans using a large number of neurons (the brain cells)? Why shouldn’t they be able to think better than humans? But then you’d have to define “better”. And how would you know if a machine was actually thinking like a human?
The standard test is to have a closed cabinet containing something you don’t know. You ask it questions via a telephone or messages. If you cannot tell from the answers whether it is a human or a computer, and it is a computer, then the computer is thinking like a human. I think computers will get there soon, if they haven’t already. A clever computer would be careful not to offer answers to a calculation that humans cannot do. If it were programmed with all the great works of music, art and literature, with every good critique of them, it could easily give human-like answers on all of them. Similarly, it could be taught everything about human emotions, sexuality and politics. I don’t think this is a very good test. You might say that this computer just had a massive memory and great skill in mimicking humans – but maybe that is sufficient to qualify as human thinking. I’m not sure.
Machines became stronger, faster, more powerful and tougher than humans a long time ago. They can work 24 hours a day; they never get ill; they could function deep under the sea or in the hottest desert or on the Moon. They have almost replaced human muscle power. Will they replace human brain power?
In the Second World War, two British geniuses with very different personalities, Alan Turing and Gordon Welshman, designed an electro-mechanical computer to decipher the German war codes. It was brilliantly successful, and played a large part in winning the war for the allies and saving countless lives. But it was large, clumsy and limited. Then in 1947 came the invention that changed the history of technology in a spectacular, utterly unprecedented way. It was the transistor – a very simple device. If you applied a voltage across it, it conducted electricity; if not, not. It was an on-off switch. It produced a number system with two numbers only: 0 and 1. It turned out to be very easy to make arrays of transistors in integrated circuits. The transistors became smaller and smaller, cheaper and cheaper, the circuits bigger and bigger, the computing power greater and greater. The Turing-Welshman computer, about the size of a small bus, could be made smaller than a pin head. The technology has advanced a trillionfold in a very short time. Machines have replaced human muscle power; now they could replace human brain power.
They have done so to a considerable extent in industry. President Trump has been complaining that America is losing jobs in car manufacturing, but he only seems aware of one of the two reasons why. It is true that imported cars have displaced some local cars in the US – mainly because they are better. But automation in US car plants might have displaced even more local jobs. First machines took over from unskilled humans; now they are taking over from skilled humans. Machine brains instruct robots how to do very precise welding, machining, fitting and painting, which they do much better than humans ever did. Automation of mining has proved more difficult than expected but I have no doubt it will come, making valuable minerals easier and cheaper to mine, saving human lives. Robot surgeons are already more accurate and precise than human surgeons. So far, the machine brains have been programmed by humans. What happens if they start programming themselves?
Computers use transistors, which are on-off switches. Animal brains use neurons, which are far more complicated, with many switches, probably switching to more than just on-off. The computing power of animals is astounding. Stationary flight, in a heavier-than-air machine, was only accomplished by a human machine, a helicopter, in 1939, and even today helicopters are cumbersome things with limited maneuverability. A humming-bird, with a tiny brain, is incomparably better. I sometimes watch swarms of midges as they fly in tight formation, sometimes hovering, sometimes rapidly changing position. The brainpower necessary for this amount of control is huge, but the whole midge is about the size of a grain of pepper; its brain must be microscopic. How do they do it? We don’t know, but that’s not to say that robots using far simpler brains might eventually produce even greater prodigies, and of course deploy enormously more power and strength.
Life on Earth began about 3.8 billion years ago. I have to say that I am an atheist but am tolerant of religions and never wish to antagonise religious people. (It is quite impossible to prove whether God exists or not. I just choose what I think is the more likely option.) I think the universe began by accident, and life on Earth began by accident, too, because the Earth happened to have the right combination of chemicals and conditions. All life now is based on the DNA molecule, which is exceedingly complex. Evolution proceeds in small steps by mutation and selection; each step must produce a viable organism. The odds against the DNA molecule evolving in one step are so astronomically high, it is essentially impossible. There must have been an earlier life – a proto-life – based on something else, and DNA must have been formed step by step under it. So there have been two life-forms so far: this unknown proto-life and ourselves, DNA-based life. Could there now be a third form: robots?
Evolution has no plan, no purpose, no aim, no intention. There are accidental mutations; most result in extinction but some result in a new viable organism. Although evolution never aims at “progress”, progress sometimes happens, and sometimes the new organism is more “advanced” than its predecessor. This is how humans have evolved from bacteria over the last three billion years. Evolution depends on one thing only: survival. The “survival of the fittest” means “the survival of those that survive” or “what works works”. The tautology is in itself a kind of proof of the theory of evolution. “Intelligent design” is a theory that all life was carefully designed by some super-intelligent being (usually meaning God). “Dumb design” is that all life evolved accidentally by some super-stupid process. God never overlooks anything because He considers everything; dumb design never overlooks anything because it doesn’t consider anything. But since dumb design only results in survival, it can produce some remarkably stupid outcomes. No naked man can look in a mirror and still believe in intelligent design. Why on Earth are those two fragile glands dangling between his legs in a thin bag of skin? Why aren’t they safely inside his body, which is where elephants keep theirs? (Human testicles need to be kept a bit cooler. Why do they need this if elephants’ testicles do not?) I can give worse examples.
Life with flesh and neurons (us) evolved to higher and higher forms by evolution, by dumb design. How would robots advance? If humans designed and programmed each robot, I believe progress would be slow but the robots would always be safely under human control. If you programmed the robots to program themselves, the progress would be much faster but there would be the danger of the robots taking over. And if you programmed each robot to program each new robot using “breeder algorithms”, which mimic the mutation and selection of evolution, progress would be faster still, and the danger greater still. But breeder algorithms, unlike evolution, have a definite aim: to optimise the design of an aeroplane wing or a combustion process or some such engineering entity. What if you gave them a different aim? If the aim was for robots to take over from humans, I feel pretty sure that it would be accomplished. What if you mimicked evolution entirely and programmed robots to make random changes in new robots and just see what happened? But now I am getting confused.
I am even more confused about death. I cannot see how higher animals, including humans, could have evolved without death: the programmed end of life for every individual animal. The parent died, the offspring lived, and so differences could be carried forward in successive generations. No death, no evolution. Would robots have to die to evolve? Would robots make new and better robots and then dismantle themselves? But what if a robot became so much like a human that it didn’t want to end its own life? I give up.
On a banal level, the Artificial Intelligence (AI) I have experienced is not only disappointing but annoying. I use Microsoft on my laptop and so have been forced to suffer “Bing” and, worse, “Copilot”. Bing is like an AI version of Wikipedia: it pretends to be offering objective truths but is actually offering woke propaganda. On every controversial subject, such as climate change or racial politics, it can be guaranteed to present the opinion of the fashionable establishment, however silly, and shut out all dissenting views. At first, I thought Copilot was a virus; it kept barging into my computer uninvited and getting in my way, stopping me from doing things. (Does anyone know how to get rid of it?) AI ideas on how to write properly are awful; they would have prevented Shakespeare from taking up literature. I use MS’s “Review” when I am checking my own writing. It is very good for picking up spelling mistakes, but very bad for offering suggestions on vocabulary and style. It tries to make you write insipid, lifeless prose without force or colour.
Back to fanciful thoughts about robots. One venture where robots would be incomparably better than humans is space travel. In fact they have already exceeded humans here. There have already been robots on Mars, sending back a mass of useful information. It is much cheaper and easier to send robots into space than humans. They don’t need oxygen or food, they can survive extreme temperatures, they can last for decades in space, and they are much better at withstanding high radiation (although some radiation might disturb transistors). So why should we want to send humans to Mars when robots can go there much more cheaply and easily and do everything there much better? This invites a bigger question: why should we want humans at all when robots can do everything so much better everywhere? I hope we don’t allow robots to ask that question.
The views of the writer are not necessarily the views of the Daily Friend or the IRR.
If you like what you have just read, support the Daily Friend.