ROBERT HENDERSON warns of the unprecedented challenges posed by Artificial Intelligence
The attention of the world is currently fixed on coronavirus, but there is another far more serious danger hurtling towards us, in the shape of Artificial Intelligence (AI) and robotics.
Both are advancing rapidly. Probably within the lifetime of most people now living – quite possibly in the next 15 years – there will be general purpose robots (GPRs) capable of doing the vast majority of the work now undertaken by humans. When that happens, international free trade and free market economics will become untenable. The real “final crisis” of capitalism will be the development of technology so advanced that it makes capitalism in the long run impossible, as machines make humans redundant across vast swathes of the economy.
Before the advent of digital technology, technological advance created new work. It may have had very painful consequences for individuals whose livelihoods disappeared – the British hand-loom weavers of the early Industrial Revolution are a classic example – but new opportunities for employment have always appeared as an economy becomes more sophisticated and variegated. The hand-loom weaver found work in the new factories – the redundant Western factory worker of today in a call centre. At worst they might only get a MacJob, but at least it was a job.
But if the GPRs can do the MacJobs as well as the more demanding work, then there will not be many new jobs for humans – not even much supervisory work because GPRs will need little supervising, and less and less of that as they become ever more sophisticated. Hence, this technological advance will be like no other; GPRs will not only take away existing jobs, they will devour any new work – the easier work first, then the more complex.
‘It’ll never happen’
The normal human response to such ideas is not reasonable scepticism, but rejection based on a refusal to accept the reality of change, a rejection expressed with ridicule along the lines of the Victorians’ response to the car: ‘It will never replace the horse’. Mention robots, and people commonly scoff ‘science fiction’ to get rid of the matter without further debate.
This type of response is natural enough because human beings, apart from disliking change, do not like to think of themselves as dispensable or redundant. Moreover, incessant propagandising by Western elites has made it received opinion that work is becoming ever more demanding and requires an increasingly educated and knowledgeable workforce – which seems to most humans to make them uniquely capable of doing the jobs of the future. By implication, this excludes mechanisation (and robots) from the majority of future human employments.
The hard truth is that most modern work requires less knowledge and skill than was required in the past. A peasant 400 years ago had to know about his soil, his plants and animals, the seasons, the weather, where natural water was, and to be able to do 101 practical things such as ploughing, sowing, harvesting, making and repairing of fences and ditches, using tools and turning out cheese and cream and dried meat and vegetables. How many jobs today require a tenth of that volume of knowledge?
Nor did more demanding work stop at peasants. A 17th century craftsman would have served a long apprenticeship. Jobs which did not require an apprenticeship would have probably required some manual skill. Those who aspired to intellectual employment had to laboriously write and amend their works rather than enjoying the immense convenience of a word processor. That, and the cost of writing materials, forced them to become precise in a way that virtually no one is today. Perhaps most importantly, modern division of labour with one person doing a repetitive job was not king. A person making something four centuries ago would probably make the entire item, and quite often a variety of other items; a 17th century blacksmith would not merely shoe horses but make a wide range of iron goods. GPRs today could take over a great deal of employment in Western economies and much of the industrialised parts of the developing world, especially China, because there are so many simple jobs which would be within the capabilities of very basic GPRs.
But that is only half of the story. If most jobs are not demanding of much by way of learned skills and even less of intellect, they do need diligence. Human beings are generally more than a little reluctant to put themselves out in work which has no intrinsic interest for them, or which is not very highly paid. So what will an employer do when he can employ a robot instead? He will go and get himself some GPRs which will do what they are told, keep working all the time without being watched, do not make regular mistakes and require no wages, or social security taxes, or holidays, or sick leave. And it will not be able to sue you for being a bad employer.
What will an employer do when he can employ a robot instead? He will go and get himself some GPRs which will do what they are told, keep working all the time without being watched, do not make regular mistakes and require no wages, or social security taxes, or holidays, or sick leave. And it will not be able to sue you for being a bad employer
In the beginning at least, there will still be a sizeable chunk of jobs which GPRs will not be able to do. These will be the jobs which cannot be reduced to quantifiable tasks – jobs which cannot be done by following an algorithm, jobs which require judgement motivation to achieve a complex end which is not obvious from the units of means which are required to achieve it. But that work is only a minority of jobs, probably a small minority, perhaps 20% of the total. If the earliest GPRs could only undertake 50% of the jobs which humans do that would be catastrophic.
There will be two further advantages enjoyed by GPRs over humans. In principle there are no limits to increases in the capabilities of GPRs; there is no such human potential in the present state of knowledge. For the foreseeable future, there is nothing to suggest that human capacity can be raised dramatically through education and training, not least because attempts to raise IQ substantially and permanently through enhanced environments have a record of unadulterated failure over the past 50 years or more.
The second advantage is that GPRs will come with a guarantee of performance. An employer gets what it says on the tin. Moreover, the performance will be consistent. Humans beings do not carry such a guarantee. The individual’s qualities only become apparent once on the job and are subject to variation according to the physical and mental wellbeing of the person. This makes them a gamble for anyone who employs them. A faulty or rogue GPR could be repaired or replaced without moral qualms; sacking a human being raises all sorts of ethical questions and matters of sentiment.
What could governments do?
When the first GPRs appear, those in political authority will probably try to say everything will be all right. It might be thought it would be pretty obvious that a GPR which could do everything the average human could do and then some would spell trouble for the human race. But it never does to underestimate the power of custom, ideology and the sheer unwillingness of human beings to face troubles which are not immediately upon them. The tired, old and worthless comparison with technological change in the past will doubtless be made, namely, that new jobs for humans will be generated by the GPRs. But that will not last long, because the reality of the situation will very rapidly force elites to accept entirely new circumstances.
There will be a dilemma for the makers and distributors of goods and services. At first it might seem attractive to use GPRs, but as humans lose their employment and the purchasing power derived from it the question for private business would be who are we producing for? Fewer and fewer people, would be the answer. For politicians, the question would be how can we finance government, including public services, when our tax base has collapsed? The answer is we cannot as things stand.
As GPRs threaten to destroy the world’s economy, politicians will be faced with an excruciating dilemma. If GPRs are allowed free rein by governments, the consequence will be a catastrophic collapse in demand as humans lose their employment en masse – highlighting the inability of the state as presently constituted to provide welfare to those put out of work or even to maintain the essential services of the minimalist state, such as the police and army.
The situation will be pressing no matter how supposedly rich a country is, because the majority of people even in the developed world are actually poor – only a few pay packets away from destitution. Even those who own their own home will not be able to sell the property because who will there be to buy it?
To begin with, attempts will probably be made to control the crisis bureaucratically by instigating rationing and price controls. But how to sustain an economy in which most people are not working? In the end, politicians will be faced with two choices: ban or at least seriously curb, the use of GPRs, or adopt a largely non-market economy. Banning GPRs completely would create a particular problem because some countries would continue to use them and this could lead not merely to cheaper goods and services but technological leaps which exceeded anything humans could do. A country which relied only on humans would be at a hopeless disadvantage.
The widespread banning of the use of GPRs in national territories would severely shrink international trade, because not all countries would stop using GPRs to produce items for export. Any country using GPRs could undercut any country which banned them. Protectionist barriers against countries using GPRs freely would have to be erected, although human nature being what it is, this would doubtless result in GPR products being supplied through a third country which had ostensibly banned GPR-produced goods and services. The likely outcome of such a situation would be for protectionism to grow beyond the banning of GPR products to the banning of products simply because they were suspected to be GPR-produced. This would also be a convenient excuse for simply banning imports.
The alternative to a protected economy in which GPRs are banned or severely restricted is a society in which the market is largely defunct. A perfectly rational and workable society could be created in which human beings stopped thinking they had to work to live, and simply lived off the products and services the GPRs produced. The GPRs would do the large majority of the work and the goods and services they provide would be given free to everyone whether or not they had formal employment. No GPRs would be allowed in private hands. Such a situation would mean the market would not make the choice of which goods and services were provided. Rather, the choice would be made by the consumer through an expression of what was needed or wanted before products were developed or supplied. This could be done through elected representatives to online voting by any member of a community for which goods and services should be supplied. For example, all available items could be voted from by the general population and those which were least popular dropped. The provision of proposed new lines or inventions could be similarly decided.
As for allocating who could have what in such a world, money could be issued equally to everyone in lieu of wages (a form of the social wage). Alternatively, in a more controlled society, vouchers or ration cards could be issued equally to everyone for specific classes of goods. Greater flexibility could be built into the system by allowing the vouchers to be swopped between individuals, for example, a voucher for footwear swapped for food vouchers.
In such societies there would be scope for a limited use of private enterprise. People could provide personal services, for example, entertainment, and produce goods just using human labour (‘human-made’ would gain the cachet ‘handmade’ has now). There would also need to be some greater reward for those who occupied those jobs which still required a human to do them such as political representation, management and administration. The reward could either be material or public approbation. It would not be unreasonable to imagine that in a society where necessary work was at a premium quite a few would take on such positions for the kudos. There could also be some legal requirement to undertake work when required.
It might be thought that the people best placed to survive would have been those in the least industrially developed states because they would be less dependent on machines. But there is scarcely a part of the world which has not been tied into the global economy. Even countries that do not manufacture products or offer services on a large scale probably export food and raw materials. One could even include the recipients of foreign aid, for that flow of money, goods, expertise and manpower is dependent on the aid-giving countries remaining economically robust.
The rate at which robotics evolves will play a large part in how the story unfolds. The speed with which GPRs replace human beings could be truly bewildering. Digital technology to date suggests that the stretch from a primitive GPR doing simple work which can be broken down into physical actions, to a GPR with some sort of consciousness or a facsimile of what humans think of as consciousness, will not be massive.
Such development could well be speeded up by GPRs assisting with development as they attain more and more sophisticated abilities. The faster the development of really sophisticated GPRs, the more chaos there is likely to be, because there will be little time to plan and implement changes or for people to accommodate themselves psychologically and sociologically.
It is reasonable to assume technology will develop until GPRs are showing behaviour which suggests consciousness. They will make decisions such as what would be the best way of achieving ends which are loosely defined, for example, an instruction to design a city redevelopment in a way which would have the greatest utility for human beings. At that point the GPRs would be effectively making value judgements.
This is a real danger with potentially catastrophic world-wide consequences. The problem is getting people in power to address the subject seriously. There needs to be discussion and planning now about how far GPRs, or indeed robots or any type, should be allowed to displace human beings in the functioning of human societies. Nor should we assume humans will happily tolerate GPRs for reasons other than economic. Robots which are too like humans make humans uncomfortable, probably because it is difficult to view a machine which looks like a human and acts like a human simply as a machine.
But the loss of jobs and incomes is only part of the problem which comes with intelligent machines. The general consequence of our ever growing reliance on digital technology is that we are increasingly being controlled by the needs of the technology, rather than using technology to serve us. It is very difficult to escape such control. A person in work will almost certainly have to use it; if in education, they definitely will. Even if a person does not encounter digital technology in their work or education, they find it increasingly difficult to avoid it in their private lives even if they refuse to use a computer or a mobile phone, because businesses and governments increasingly require those dealing with them do so by computer. People are being driven to own and use computers to avoid feeling isolated and excluded.
Despite all these pressures, there are still a large number of people in Britain who have remained distant from the digital world. According to a 2019 Office for National Statistics report, millions of British adults have never been online. It is unreasonable in a civilised society to simply hang the computer-ignorant or the intellectually-underpowered out to dry as digital technology looms ever larger. Yet that is precisely what is happening.
There is one thing the government of any advanced country should do – create circumstances in which those who cannot come to terms with digital technology can live in an ever more computer-controlled world. They can do this by maintaining non-computer access to state-funded organisations and forcing through legislation larger businesses and not-for-profit organisations to do the same. Worryingly, there is little evidence that UK politicians are taking this problem seriously. There have been rather half hearted attempts to ensure that cash point machines are provided so that no one has to travel more than a few miles to draw cash, but that is wholly inadequate because many people, especially the old, cannot readily travel several miles.
At the same time the UK government is dragging its feet over making access to cash a legal right. Failure to do so could all too easily allow the UK to sleepwalk into a cashless society, a state of affairs which would not only potentially give the government immense opportunity to intrude on private lives, but be a constant worry even for those accustomed to digital technology.
Someone living in Britain between 1815 and 1914 saw more radical technological change than any generation before. But that change was the difference between living in a still largely pre-industrial society (in 1815) and an industrial society in its early middle age (in 1914). Moreover, the change did not require the vast majority of the population to master complicated machines at their work, let alone in their own homes. In 1914 the most complicated machine most people would have had to operate was probably the telephone, and vast swathes of the population would not even have had to go that far into the world of technology.
In the past 30 years, all this has changed hugely. We are now in a world in which computers are absolutely integral to business and public administration, and are the norm rather than the exception in homes. For most people, it is literally impossible to escape them. Worse, they have become ever more complex and demanding to use, and invade ever more of our lives, as microprocessors are inserted into the most unlikely things, such as clothes. All machines are becoming more and more demanding. We desperately need politicians who will act to avert the looming disaster this unique situation threatens to bring. Don’t hold your breath waiting.
ROBERT HENDERSON blogs at Living in a madhouse and England Calling. He is also on Twitter @RH156