Artificial Intelligence: Some Issues.
By Mark Cowling
Following its disastrous performance at the last general election, Labour is naturally having an extensive rethink of its policies. Some of this is reproduced in Labour Affairs 310 (August 2020). Artificial intelligence is having increasingly pervasive effects, and needs to be considered as part of the background against which policies are developed. In this article I want to reflect on the potential of artificial intelligence for human liberation, but also on its darker side: its acceleration of the tendencies of neoliberalism, and in particular of class polarisation and its potential for the surveillance of minorities. What role might it play in a socialist future? Labour must grasp both the threats and the opportunities offered by artificial intelligence.
The Rise of Artificial Intelligence.
There is a substantial and increasing volume of literature on artificial intelligence.
The Oxford Dictionary definition of artificial intelligence is: theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages. Artificial intelligence works by gathering huge quantities of data and then analysing it. According to Smith and Browne, we are ending this decade with about 25 times as much data as at the beginning of the decade.
The quantity of data that tech companies hold on individuals is enormous. One Viennese citizen eventually prised out of Facebook a CD-ROM containing 1200 pages of data. He had “consented” to this by using Facebook. Something similar happens when people use websites. These invariably want to place a cookie on your computer. You have “consented” to this by allowing websites to place cookies on your computer. This allows them to harvest some of your data. Prior to 2016 this would also involve “consenting” to other third-party providers harvesting your data. A calculation in 2008 suggested that it would take 76 full days a year to read through all the agreements. This will have grown considerably since that time. A team of researchers found in 2015 that a visitor to the 100 most popular websites would accumulate no less than 6000 cookies, 83% of unrelated to the initial website visited. This type of excuse for harvesting virtually unlimited quantities of data led the European Union to develop its General Data Protection Regulation in 2016. This places some limitations on what people are actually consenting to. US citizens do not enjoy this protection. This leads Zuboff to describe the current era as one of Surveillance Capitalism: ‘a rogue mutation of capitalism, involving concentrations of wealth, knowledge and power unprecedented in human history; the foundational framework of a surveillance economy
;… A threat to human nature’. This is by no means the full definition, but it gives the general flavour of her book. The basis of Google’s massive accumulation of wealth is its unprecedented intrusion into personal life. With the ability to probe what people are doing online, using cheap cameras to photograph geographic locations, and cheap storage, ‘Your whole life will be searchable’ as Larry Page, one of the founders of Google, remarked in 2001, with no concern about how this would be intruding into personal life. Google and Facebook lobbied vigorously to oppose any restrictions on their right to intrude, Google spending around $18 million each year on lobbying in Washington, and it is also the largest individual lobbyist in the EU. Another intrusion fostered by Google is the ability to track one’s location via one’s smart phone, all for the benefit of advertisers – thousands of locations can be tracked each day. Currently with the Covid-19 pandemic this is a particularly delicate issue. There is the possibility of using smart phones as a way of tracing contacts who may have been infected with the virus. Indeed, both they and the person who infected them may actually be asymptomatic, but nonetheless have the possibility of infecting others with consequences which may even be fatal. However, intruding into people’s lives in this way is also potentially a serious violation of their civil liberty. The British government is currently attempting to enhance the rather dismal performance of its “world beating” (Boris Johnson) contact tracing system with an app jointly developed by Apple and Google.
This capacity for intrusion can have positively sinister results. In June 2013 the Guardian revealed that nine major
American companies which handle data, including Apple and Microsoft, had signed up for a program called PRISM, which allowed the NSA (the US National Security Agency) to spy at will on anyone who made use of programs provided by these companies. The Guardian knew about this thanks to Edward Snowden, who absconded from an NSA centre bearing about a million documents, which kept journalists busy for the next year or so. Intrusion by the NSA was prima facie a violation of the Fourth Amendment, which guarantees US citizens security from unwarranted searches and seizures, and, obviously, there are similar expectations in other Western democracies.
A variety of dilemmas for tech companies arise: some intrusions by security services are legitimate, for example attempts to locate extremists planning terrorist incidents; should they be disrupting anti-vaccination groups? Should they be disrupting the glorification of terrorism? Can they distinguish that from legitimate discussion of terrorism? Likewise, for websites encouraging sex with children, which are fostering serious lawbreaking, but need to be distinguished from academic discussions of paedophilia.
The capabilities of artificial intelligence are developing extremely rapidly. One impressive recent example was the victory of a computer against the world champion of the game of Go, which is particularly complex and difficult to get a computer to deal with competently. Spectators were particularly impressed with a move made by the computer which made no apparent sense to them, but which turned out to be a stroke of genius, and was effectively the program being creative. All kinds of beneficial applications keep emerging. Three examples in the field of medicine: there is a serious and increasing problem of bacteria which are resistant to antibiotics, but a computer program was set to work and looked at less likely possibilities with the result that a superior antibiotic was discovered in an unlikely place; computers can now check the results of x-rays and scans in some cases more reliably than doctors; robot assisted surgery is developing rapidly. The modelling of the climate and its development presents extremely complex problems, but scientists are getting better at this task, in part thanks to artificial intelligence. Another area in which there are beneficial developments helped by artificial intelligence is agriculture. Tractors can now plough or spray very precisely thanks to GPS positioning, reducing the tendency for sprays to go further than their targets. Indeed, tractors could soon be self-driving as well. There is an ongoing problem in Britain of getting hold of workers to do the hard labour of harvesting fruit and vegetables, and if artificial intelligence would get some of this work done by robots it would be helpful.
Obviously, however, there is a darker side. Staying with agriculture, particularly in the United States, huge numbers of animals such as pigs are basically kept remotely by computer in conditions which pay little attention to their welfare. Pigs are very intelligent animals, and it is particularly cruel to do this to creatures capable of playing simple computer games. In the military field, drones are operated using artificial intelligence which can go wrong, with the result that innocent people are killed, which is terrible in itself and which thwarts the military objectives. The demands of the computer industry can lead to pollution and using up scarce resources.
A new stage of capitalism?
We are entering a new phase of capitalism. Marx identified stages such as cottage industry, in which merchants go round to cottages in which, for example, weaving is being carried out; they bring with them fresh supplies of wool and take away woven cloth, making their profit from the labour of the weavers. Next comes association, in which the weavers are gathered together in one place; then machinery, in which more machinery is introduced; then manufacturing, in which there is the application of steam or electric power; as this develops there is a system of machines. This is about as far as things got in Marx’s lifetime. Another major development was an assembly line on the model developed by Taylor and pioneered in the Ford motor factory. Next, with the increasing development of computers, comes post-Fordism in which rather than “any colour you like, so long as it’s black” is replaced by a much more diverse set of possibilities, with very many models and options of automobile being available. There is also much more diversity in every sphere, so that with single keystroke input publishing a small new magazine becomes possible much more easily, leading to many new titles becoming available. All sorts of new enterprises can be started more easily, as viewers of Dragon’s Den will be well aware.
Automation has advanced very considerably. A viewing of Inside the Factory with Gregg Wallace rapidly makes this apparent. Assembly lines appear only rarely. Instead the workers are largely supervisors of a system of machines. Their role is quality control, and moving the product from one process to another. Just to take two examples among many: Heinz baked beans are made with haricot beans imported from the United States. Some of these are perfectly nutritious but discoloured. This used to be dealt with by people with spoons scooping up unsatisfactory beans. They are now detected by laser and dispatched for animal feed by puff of air. Walkers crisps are made from potatoes with a lower sugar content than standard potatoes. This is to stop them becoming discoloured. In spite of this starting point, some crisps are still too dark. Each crisp gets photographed, and discoloured crisps are disposed of automatically. When it comes to warehouse operations, human beings are not needed inside some warehouses; everything is controlled from outside.
The current degree of automation enables very large quantities of products to be the work of relatively few people. Thus, all the many Heinz products made in Britain, including 3 million cans of baked beans, in a factory which works round the clock and seven days a week, require a workforce of about 1500. This explains why, although the UK remains the seventh-largest country for manufacturing in the world, the manufacturing workforce is much smaller than it used to be. Between the early 1980s and 2018 the manufacturing workforce shrank by 3 million, representing a fall from 21% of the UK workforce to 8%. Despite this, from 1970 to the present day the country has ranked somewhere between sixth and eighth in the world, with China now top, the United States second, Japan third, Germany fourth, South Korea fifth, and then France, Britain, Italy and various other countries moving in and out of 6th to 9th place. Bastani points to similar effects of artificial intelligence and the use of robots in US industry.
The increasing application of artificial intelligence is likely to result in increasingly automated operations. Thinking now of distribution, in an Amazon warehouse the workforce follows instructions from a computer about where items which form part of a customer’s order can be found. Jeff Bezos would obviously like to increase his wealth, so it would be helpful if a robot could do the selection and wrapping currently done by people. It could then load the parcels onto a self-driving lorry and move them on to their destination. Even better, maybe a drone could deliver the parcels. We are rapidly moving from post-Fordism with automation to a stage of artificial intelligence.
While this stage obviously offers the potential for an unprecedented degree of human liberation, at the moment there is a sharp increase in polarisation, with the massive accumulation of wealth by the tech firms at one pole and a growing army of people stuck on minimum wage at the other.
Moreover, Americans (and almost certainly the same is true of British people) trust their government less and less. In 2017 only 18% of Americans trusted their government. Now that Trump has dealt so very badly with the coronavirus crisis, there is every reason to think that those who still remain alive trust the government even less. Mike Pompeo, US Secretary of State, visiting London with a view to stirring up anti-Chinese and anti-Russian sentiments, said amongst other things that the United States government did not go about assassinating people! Numerous public figures in Latin America and innocent victims of drone strikes demonstrate that this was a blatant lie, as a viewing of the film Bowling for Columbine reminds us: the film includes a section which lists the countries in which the United States has intervened in order to prop up dictatorships and remove figures who opposed the US interests. The role of the CIA in replacing elected president Allende in Chile and replacing him with General Pinochet is a particularly striking example. The United States may be a democracy, but it is one which has committed extensive state crimes. When leaders tell blatant lies of this sort it is hardly surprising that public trust is so low.
Artificial Intelligence and Employment/Unemployment.
Previous technological developments have led to a change in employment patterns, which can be extremely unpleasant for the individuals concerned. In the nineteenth century many people were forced off the land and ended up in noisy, unsafe factories working long hours and living in bad conditions, as charted so brilliantly in Capital. A major disruption is extremely likely thanks to artificial intelligence. Call centres are likely to require far fewer staff, as voice recognition software improves and enquiries can be dealt with by computer programs. Kessler charts how, since around 2002, the bulk of the growth in the US employment market is in the form of self-employment. People are increasingly independent contractors, and therefore do not get the fringe benefits (holiday pay, sick pay, medical cover, pensions etc.) enjoyed by employees. Whilst highly skilled professionals, notably programmers, website designers etc. can do very well as independent contractors, these and similar professionals actually employed by the tech firms can also earn spectacular amounts, they are a tiny portion of the workforce, whereas for unskilled workers such as cleaners and janitors, not only is there the problem of fringe benefits mentioned above, but overall earnings are typically lower than those of cleaners lucky enough to be employees. At its worst, the gig economy produces conditions approaching those of debt slaves, who found that once actual slavery had ended, they became bound to the former slave owner as a result of indebtedness. Kessler found examples of people who were in something resembling this condition as a result of purchasing their vehicles with assistance from Uber. Strikes, the traditional weapon of the working class, have become increasingly ineffective: in the United States they have diminished from around 300 per year in the 1950s to around 20 per year since 2000. It is little wonder: Americans are desperately insecure. In 2015 a report by the Federal Reserve found that 47% of Americans could not cover an unexpected expense of $400 from their savings or their credit card. They are likely to have inadequate medical insurance, so such an emergency is only too possible.
Another dark side of this new stage of capitalism is that some of what it depends on is linked to highly toxic labour, notably the manufacture of silicon chips and lithium batteries. A different form of toxic labour, much of it performed in the Philippines, is an army of people who spend their days working out whether digital content is so obnoxious and pernicious that it needs to be eliminated. Dyer Witherford says that the number involved is 100,000, but there is every reason to think that this will have grown considerably since. Or what about the by now well-known situation that prevailed at the plant in China where iPhones are manufactured, where conditions are so oppressive and regulated that many workers were driven to suicide?
In the foreseeable future an extensive disruption can be foreseen as very large numbers of workers are displaced by artificial intelligence. Thus factories will be increasingly staffed by robots; shelves will be stacked and clothes folded by robots.; drivers will be replaced by self-driving vehicles – according to Russell cars in cities will largely be replaced by free self-driving buses; perhaps the poor people forced to watch a daily diet of extreme pornography will be replaced by computer programs; low-level legal work will also be computerised – computers did better than law professors at analysing non-disclosure agreements; insurance underwriting will soon largely be done by computer, as will at least some medical diagnoses, as will telemarketing, credit checking, tax accountancy, operating checkouts and baking.
In contrast to the accumulation of poverty at the bottom end of society, there is a massive accumulation of wealth at the top. Russell reproduces an alarming graph produced originally by the US Bureau of Labor which shows from around 1970 onwards productivity doubling by 2005 while the rewards going to workers in the sector which produces goods remained static. As Susskind points out, Amazon, Google and Facebook have an accumulation of wealth roughly equivalent to the GDP of Canada. There are structural reasons why this is so: these firms possess an oligopoly of data, making them hard to displace; and they have relatively few employees, meaning that the number of people who benefit from their position is relatively small. As factories become automated, work previously taken offshore to, for example, China, will increasingly be done by robots in, for example, Germany. Plainly, extremely dramatic disruptions are in prospect. At least some of the people whose lives are disrupted will turn to crime, and doubtless their crimes will be detected using artificial intelligence. The socialist solution to this may well be a universal basic income, which will be briefly discussed below.
Artificial intelligence, Crime and Policing.
The Internet plays a major role in propagating and facilitating terrorism. ISIS does a great deal of its recruiting and basic training by posting appropriate videos online. The horrific attack by an Australian in New Zealand on March 6, 2019, in which 50 innocent Muslims were killed, was doubtless originally inspired via the Internet, and also was publicised on the Internet. This obviously poses major problems for security and intelligence services. They have a legitimate interest in trying to prevent horrific crimes of this sort. This involves trying to remove material glorifying terrorism from the Internet, and they will also naturally want to try to find the authors of such material. But there are also dangers to civil liberty. What about people who are simply curious and have no intention whatsoever of engaging in terrorism? What about academic researchers? How do you distinguish a group of academic researchers, who are posting material produced by terrorists with a view to discussing it, from a group of actual terrorists? Obviously similar issues arise with other forms of criminal activity, for example fraud, paedophilia, or posting death threats.
The above difficulties are, however, a mere foretaste of things to come! In 2013 Harvard researchers managed to intervene in the brain of a volunteer in such a way that he was able to control the tail of a rat with his thoughts. Other experiments in which one volunteer was able to control the hand of a second volunteer on the other side of a university campus simply with his thoughts, and, in a different experiment a volunteer was able to control the movements of a cockroach with his thoughts. Apart from the interesting idea of something going wrong and the cockroach getting in control of the volunteer, a scenario worthy of Kafka, in the future it might be possible for ISIS to turn captives into suicide bombers controlled by the thoughts of an ISIS leader.
Another possibility is the use of transcranial Direct Current Stimulation (tDCS), which is definitely able to increase alertness and cognitive performance, but it can also stimulate or inhibit emotional responses. This obviously raises all kinds of possibilities. A particularly concerning one is that our emotional responses are an important part of how we recognise particular environments as inviting or threatening. This in turn could be adapted for autonomous weapons systems, meaning battlefield weapons with a sense of the nature of their environment. Apart from the potential to kill the wrong people by accident, autonomous weapons systems could also get left around after the end of a particular conflict, and go on killing people long after the conflict was over, rather on the model of the 110 million or so landmines estimated to be still in the ground. Yonck agrees with a campaign underway to ban the use of autonomous weapons systems before they are properly developed. The campaign is, in fact, reminiscent of Robert Oppenheimer, who organised the Manhattan project which developed the atom bomb, proposing some form of international control (for a much more detailed account of the need to limit the use of autonomous weapons see Tegmark).
Affective computing is potentially also extremely useful for cyber criminals. If a computer program can recognise and manipulate an emotional state, this can be used to facilitate all kinds of scams. People as sophisticated as the former editor of Psychology Today, an expert in affective computing, can be tricked into corresponding with a chat bot; according to an estimate in 2014 about half of Internet traffic is generated by bots, and some 30% of these bots are malicious, with fully 20% of Internet traffic being generated by impersonator bots. Another unpleasant possibility is criminal intervention in the running of pacemakers, internally worn defibrillators, externally worn insulin pumps and similar devices. The wrong sort of intervention has generally not been guarded against by manufacturers, but would obviously be potentially rapidly fatal, making blackmail a serious possibility.Affective computing can also be misused by the police in the course of interrogations. A suspect can be induced to “remember” things which did not actually happen.
The role of affective computing in recent elections is worth a study in its own right. The team which got Trump elected in 2016 made extensive use of very carefully targeted newsfeeds and advertisements. At least some of these appear to have originated from Russia, which has an interest in keeping the West divided, for which purpose Trump’s vigorous America First policies serve very well. The same issues arise for the Brexit referendum, both in terms of the skilled use of affective computing by the leave camp and apparent Russian intervention in favour of Brexit. The tech firms have been trying to tighten up on this issue in time for the 2020 presidential election, and other elections elsewhere. None of this is to suggest that extensive emotional manipulation using artificial intelligence is the only thing which wins the popular vote. However, given that Hillary Clinton actually got more votes than Trump, and that the Brexit referendum would have gone the other way on a swing of 3%, the role of affective computing may have been enough to swing both votes. Artificial intelligence has thus almost certainly facilitated serious crimes against the democratic process.
A particular issue in the area of policing is the role of facial recognition technology. Facebook is devoting very considerable effort in this area, and is able to identify mood, gaze, gait, hairstyle, clothing, activities, interests, body type and posture. Apple are also interested in facial recognition technology, and the more recent iPhones are unlocked by facial recognition. Payments can also be made by facial recognition. Apple are therefore very confident about their technology, although it may have difficulty with identical twins. Facebook’s intention in analysing biometric markers is to modify human behaviour for the benefit of advertisers, but this ability can also be used by states to modify the behaviour of citizens. Back in the 1970s psychologists debated the ethics of behaviour modification and came up with a code of ethical practice. Zuboff wants something similar for programmers of facial recognition technology. A full discussion of the use made by the Chinese state of facial recognition technology would extend this article excessively, but the current intention is plainly to control the life of Chinese citizens in considerable detail, and without most of the restraints which are found in liberal democracies. A good example of a restraint on police power achieved in this area in Britain came in August 2020. Ed Bridges, a citizen of South Wales, objected to the way in which the South Wales police had collected his biometric personal data, once when he was Christmas shopping and once when he was on a peaceful demonstration. The South Wales police maintain secretive watchlists, and, once they have collected the data of anyone who comes in range of their cameras is then compared with a watchlist. The Court of Appeal agreed with Mr Bridges, whose case had been taken up by Liberty, a charity whose purpose is to maintain civil liberties. South Wales police and other British police forces will now have to seriously rethink their use of facial recognition technology. Liberty note that several US cities have banned the use of facial recognition technology, and argue that we should do the same in Britain.
Ferguson provides a thorough account of the use of artificial intelligence for policing purposes based on a unit of the Los Angeles Police Department (LAPD) called the RACR or Real-Time Critical Analysis and Response Division. Roll call for police officers is a bit different from what one would expect from the television. They are provided with a digital map of their area, complete with a crime forecast which shows the city blocks where the algorithmic analysis of large quantities of data, harvested from some of the 4000 databases which monitor everybody’s behaviour, suggest that there is likely to be trouble. When they are on patrol reports of a gang fight result in the officers getting data about what they are likely to find sent to their mobile phones: the history of gang activity in the area, likely tensions between gangs, and updates on how things are progressing, so that they know what to expect when they arrive. The possible downside of this is that it may just be giving a scientific gloss to existing prejudices, particularly racial prejudices, reinforcing existing bias, fear and distrust. Moreover, in the United States at least, there is invariably an exemption for law enforcement officers from any restrictions on the access to data or to its dissemination. Putting big data at the service of grossly biased, incompetent, oppressive, revenue driven police departments such as the one in Ferguson Missouri is likely to make bad policing more efficiently bad! Data driven policing is one response to the outrage felt in black communities over numerous police killings of unarmed African-Americans. But there is also the possibility of monitoring and predicting police misconduct, and analysing the social and environmental roots of criminal behaviour.
There is theory that criminal behaviour is located in particular networks, and this can be quite useful if it is combined with the analysis of data using artificial intelligence. For example, in one analysis of the situation in Boston, belonging to a small network comprising 4% of the city’s population raised one’s chances of being murdered by 900%. An obvious observation about US murder statistics is that the relatively high murder rate of nearly 5 per hundred thousand of population is quite a lot higher than that in the UK with its 1.2 per hundred thousand (Wikipedia). One obvious explanation is the unfortunate US habit of people shooting each other, although there are doubtless a whole series of other factors. A lot of Canadians possess rifles for shooting game, but Canadians murder each other at about a third of the rate of the United States. Much of what follows concerns negative aspects of big data policing, but it does have useful potential, notably in identifying forced migration for purposes of sexual exploitation, and men liable to commit domestic violence.
Ferguson is concerned that big data policing may simply reflect existing biases. For example, black and white populations imbibe marijuana at much the same rate, but arrests for marijuana possession are overwhelmingly of black people. This is particularly heartbreaking now that an increasing number of US states have decided that marijuana is enjoyable rather than criminal. Similarly, imprisonment rates are notoriously skewed with far more blacks in prison than would be expected from their numbers in the population. Presumably part of the explanation is that this is a legacy from the days of slavery. However, similar phenomena are found in Britain, with more black people than would be warranted from their numbers subjected to being stopped and searched, having their car stopped, or ending up in prison. Moreover, if the police focus on any part of the population – for example, members of the clergy – they are bound to discover more criminality in that group. The assumption up to now is that the database is reasonably accurate, but this is by no means always the case, and plainly if the data put into it is wrong, for example it identifies infants as gang members, the results are going to be garbage as well. Moreover, because police databases are held in a secretive way, it is extremely hard to challenge wrongful inclusion in one.
If the database focuses on black areas, this may in turn justify aggressive policing – in Britain and excessive use of stop and search, which is simply a feature of everyday life for young black men in some areas. Even more alarming is the aerial surveillance by the FBI of peaceful protests in the wake of the shooting of black people – essentially the immediate precursors of the current Black Lives Matter. Of course, things have been exacerbated by the way in which aggressive policing and white vigilante actions have been encouraged in the current round of protests by President Trump. Indeed, the very detailed level of information available to government can feel like something very reminiscent of Big Brother. Data is collected from police officers in a manner which is prone to vicious circles. For example, just being stopped becomes an additional reason for them stopping you next time, and if you are on the Chronic Violent Offender list you are likely to be stopped in an aggressive manner, making it more likely you will object, just as a violent offender would. This can have very serious results, notably in the case of the shooting of Mark Duggan, which triggered off the riots of 2011. There is also a class dimension to these issues. Young people of all races and classes get drunk, use drugs and engage in minor criminality, but those at the top of the social scale do not generally get their collars felt. Indeed, if they are British and members of the Bullington Club they end up in the Cabinet. Some officers are responsible for much more than their share of incidents where they use excessive force, and a more beneficial use for big data is to identify them and retrain them.
Ferguson makes many useful observations, but says nothing about crimes of the powerful, which a typical police patrol does not encounter. Certainly not state crime; not the sort of corporate crime which produces pollution or seriously faulty goods; not fraud including, of course, increasingly sophisticated Internet fraud; not criminally unsafe working conditions, nor dangerously long working hours with insufficient breaks for rest and refreshment.
A socialist future?
How could artificial intelligence contribute to a socialist future? How could it contribute to getting there?
While Wikipedia and associated wikis could hardly be considered socialist enterprises, a collective enterprise of this sort, partially automated, and not for profit, could certainly feature as part of a socialist society. Similarly, as Susskind points out, the use of Creative Commons licences allows the sharing of knowledge freely on an unprecedented scale, with over a billion documents included. The authorities can increasingly monitor traffic, air pollution, utility usage, lighting and so forth. Obviously, this can take on a Big Brother aspect, but it has also the possibility of fine-tuning traffic lights or working out ways of reducing air pollution in previously unprecedented but beneficial detail. Protests can be facilitated by the new media, including Occupy and the Arab Spring (and, indeed, the UK riots of 2011).
Another interesting possibility is the – generally anti-authoritarian – use of political hacking, or more prosaically the vigorous use of freedom of information legislation. Susskind identifies some threats to democracy. The first of these is what he calls perception control: if you are getting your news from an online source, then the control of the content and bias of the news gives considerable power to whoever is exercising it. Moreover, political reality becomes increasingly fragmented. Susskind gives the example of news originating from Twitter: if you are a Republican 90% of the tweets you receive will be from Republican sources; if you are a Democrat 90% of your tweets will come from Democrat sources. Added to this, particularly facilitated by Donald Trump, is the debasement of the term fake news, which has come to mean, for him at least, anything you disagree with. Thus, also the idea that there is now post-truth politics, which in turn leads to a fragmented reality in which your group effectively lives in a different world from mine; and this is exacerbated still worse by the very precise tailoring of political messages to increasingly narrow groups of people. Moreover, all the previous problems of distortion are exacerbated still further by the existence of bots, largely supporting Trump in the USA, and generating a third of tweets in the run-up to the EU referendum in the UK, almost all of them pro-leave. Similar observations about the debasement of public life may be found in Foer: he notes that journalists tend now to relentlessly pursue clicks on their publication’s website rather than an important story for its own sake. On the other hand, he notes an encouraging trend in the opposite direction: the New York Times markets itself as a responsible publication in contrast to the debased behaviour of Facebook. In the immediate aftermath of Trump’s electoral victory, it gained 130,000 new subscribers. Subscriptions to the Guardian have very similar motivation. In marked contrast, Facebook actively boasts about its ability to manipulate behaviour. The Covid-19 crisis has been a corrective to this: political leaders cannot simply invent reality to suit themselves, as the existence of large numbers of sick and dead Americans demonstrates. As a result of his inept handling of the crisis, public levels of trust in Trump are at a low ebb, in contrast to, for example, Angela Merkel, who is held in high esteem because her handling of the crisis has been based on sound scientific evidence. Likewise the very substantial Conservative lead at the time of the election has eroded to nothing as a result of inept handling of the Covid 19 crisis.
Susskind belongs to the broad liberal tradition, but some of the issues he considers would also be of interest in building a socialist society, such as what should be done about personal good or bad luck, or the desire for recognition of the kind being asserted, for example, by the Black Lives Matter movement. he makes a call for transparency in the construction of these algorithms so that it can be challenged, and also for the tech firms to provide a straightforward explanation of what they are doing. There is also, he argues, a need to break up the extreme concentration of power in the hands of the tech firms. while socialists would surely want the wealth of the tech firms applied for the general good of society rather than just as their owners see fit, problems of undue concentration of power do not simply go away in a socialist society. There is a nice illustration of this from Foer. He talks about almost accidentally starting an uneven battle with Amazon, and finding a whole variety of academics unwilling to fall out with them because they are afraid that evil things will happen to their books on the Amazon website. One is reminded of authors wondering what they could get away with in the Soviet Union. Although, thankfully, Jeff Bezos does not, as yet, have access to the Gulag.
Foer identifies a very important issue which is frequently written about in the Guardian, namely the avoidance/evasion of tax by the tech firms, by employing every device known to tax accountants and lawyers, and creatively inventing new ones, thus paying far less tax than the likes of Walmart.
Marx and Engels identified figures such as Robert Owen, Fourier and Saint Simon as utopian socialists who sketched out what a socialist society might look like, in the case of Fourier, particularly, in delightful detail, but who failed to provide the economic and social analysis of existing society and any realistic political means of getting from here to there. Bastani provides something rather like utopian socialism for today, sketching out possibilities based on artificial intelligence and science which is available now or which could easily be developed in the near future, but without really dealing with the issue of how to get from here to there. He identifies five crises: climate change; scarcity of some resources, notably energy, water, and some minerals; an ageing population in the advanced countries; a growth in the number of poor people and most important of all, increased unemployment based on the rise of artificial intelligence. The banking crisis of 2008 demonstrated that there is socialism for the rich, where failed bankers get bailed out, while poverty and the use of food banks at the other pole has grown, and, in the UK, by 2016 60% of households in poverty contain somebody in employment. Yet there is enough solar energy available to easily meet the demands of humanity and, when the problems of mining on the near-earth asteroid Psyche are solved, there will be no scarcity at all of a variety of minerals. With an abundance of computing power, of energy and of minerals, there will be much less need to work at mundane tasks, leaving people free to be creative in a luxurious society. If this is basically Marx’s vision of communism, strikingly Keynes had the same view of a society beyond scarcity. Intriguingly, Russell suggests that much of this might take the form of creating in virtual reality. Fuchs points out that Marx in the Grundrisse notes that there is a contradiction between the increasing productivity of labour on the one hand, making a radical reduction of working time possible, and value and surplus value being based upon labour, and also introduces the notion of general social knowledge becoming a direct force of production and increasingly controlling social life. A vision of a society of plenty can be gained from what has happened in agriculture. Throughout history, most of humanity has made its living by working on the land, yet in Britain, France, Italy and the United States a tiny portion of the population, between 1% and 4%, makes its living on the land, something which would have seemed impossible in the past. Rising world populations pose a problem when it comes to meat production, which could in principle be easily solved by the production of synthetic meat, with the use of far less natural resources than required to produce meat today, particularly beef. Alternatively, of course, everybody could move on to something close to a vegan diet without the need for any scientific advances.
Bastani argues that the changes required by humanity’s situation render neoliberal politics, which has brought us the banking crisis, the Grenfell disaster, the Carillion collapse and overpriced privatised rail companies in Britain, and similar phenomena elsewhere, outdated: a collectivist approach is needed. He sees unions devoting some of the money in their pension funds to creating useful jobs and facilities in Britain rather than seeking profits abroad. The objective should be that more and more important basic services such as public transport should simply be free on the model of bus passes. Bastani recognises that he is only sketching the outlines of a socialist future, and certainly does not think that the changes he wishes to see would represent an endpoint in human history; they are rather a beginning. While this presents some sensible ideas, I describe what he offers as utopian socialist because there is no real overall account of how you might get from here to there even in Britain let alone elsewhere.
Srnicek and Williams try to address this question. They argue that the decline of social democratic politics and the growth of what they describe as folk politics, meaning a politics which concentrates on specific issues, which comes and goes quickly in the way that, for example, the various protests about aspects of globalisation, or the Occupy movements did, means that people on the left tend not to look at global and abstract issues of the kind which need to be addressed in order to bring about significant change. The victory of neoliberalism has led to a disillusionment with electoral politics, and reduced participation rates. What is needed, they argue, is an aspiration for the flourishing of all humanity. This involves the traditional social democratic provision of basic needs, the expansion of social resources, and the development of technological capacities. But this is a form of humanism without a pre-established end point. The left, they argue, should mobilise around a post-work consensus. They note that between 47 and 80% of existing jobs are capable of being automated out of existence. They call for a universal basic income, which would both help with this situation, and actually act as an incentive for further automation because wages for the worst paid jobs would rise. They seek to establish their view as a new common sense. They also argue that very many of the innovations made by the tech firms are actually based upon government investment, so that there is a legitimate interest in subjecting them to public control.
Although, given its date, their book obviously fails to take account of Corbyn’s revival of the Labour Party and its subsequent electoral disaster, it does offer a vision of a socialist way forward.
These articles have demonstrated that artificial intelligence is extremely important for the understanding of how society is going to develop over the next few decades. It shows that there are both serious threats, such as the potential for military uses, surveillance of whole populations, and electoral manipulation. But there is also a massive potential for moving society forwards in a socialist direction. If Labour can get to grips with what is involved it will be well placed to take advantage of the very poor showing that the government is making.
This is from Problems 43, October 2020. Other issues at https://labouraffairsmagazine.com/problems-magazine-past-issues/. And the whole magazine is available as a PDF at https://labouraffairsmagazine.files.wordpress.com/2020/11/problems-43-tragedy-and-ai.pdf
 Adams, J. and Kletter, R., Artificial Intelligence: Confronting the Revolution, Endeavour Media, 2018; Bastani, A., Fully Automated Luxury Communism: A Manifesto, London: Verso, 2019; Dyer-Witherford, N., Cyber-Proletariat: Global Labour in the Digital Vortex, London: Pluto Press, 2015; Dyer-Witherford, N., Kjosen, A., Steinhoff, J., Inhuman Power: Artificial Intelligence and the Future of Capitalism, London: Pluto Press, 2019; Ferguson, A. G., The Rise of Big Data Policing: Surveillance, Race, and the Future of Law Enforcement, New York University Press, 2017; Foer, F., World without Mind: Why Google, Amazon, Facebook and Apple Threaten our Future, London: Vintage Digital Publishing, 2017; Fuchs, C., Rereading Marx in the Age of Digital Capitalism, London: Pluto Press, 2019; Kessler, S., Gigged: the Gig Economy, the End of the Job, and the Future of Work, New York: Random House Business, Second Edition, 2019; Russell, S., Human Compatible: Artificial Intelligence and the Problem of Control, Harmondsworth: Penguin, 2019; Schwab, K. and Davis, N., Shaping the Fourth Industrial Revolution, World Economic Forum, 2018; Smith, B. L. and Browne, C. A., Tools and Weapons: the Promise and Perils of the Digital Age, London: Hodder and Stoughton, 2019; Srnicek, N. and Williams, A., Inventing the Future: Post-capitalism and a World without Work, London: Verso, 2016; Susskind, J., Future Politics: Living Together in a World Transformed by Tech, Oxford: Oxford University Press, 2018; Tegmark, M., Life 3.0: Being Human in the Age of Artificial Intelligence, Harmondsworth: Penguin, 2018; Yonck, R., Heart of the Machine: Our Future in a World of Artificial Emotional Intelligence, New York: Arcade Publishing, 2020.; Zuboff, S., The Age of Surveillance Capitalism: the Fight for a Human Future at the New Frontier of Power, London: Profile Books, 2019.
 2019, location 114.
 Smith and Browne, 2019, location 2009.
 Zuboff, 2019, location 919.
 2019, location 69.
 Zuboff, 2019, location 1784.
 ibid., Location 2291.
 ibid., Location 4403.
 Smith and Browne, location 250.
 see, for example, Tegmark, 2018.
 Schwab and Davis, 2018, have numerous further examples of potentially beneficial applications of artificial intelligence.
 2019, location 1001.
 For a more detailed analysis, with onward references see Tegmark, 2018, p. 119.
 Zuboff, 2019, location 6945.
 Smith and Browne, 2019, location 3548.
 2019, p. 9.
 ibid., p. 89.
 ibid., p. 106.
 ibid., p. 159.
 ibid., p. 189.
 Dyer Witherford, 2015, location 2105.
 ibid., Location 2433.
 ibid., Location 3065.
 Russell, 2019, p. 74.
 ibid., p. 67.
 see, amongst others, Dyer Witherford 2015, location, 3570, Bastani, 2019, location 1200, Russell, 2019, pp. 118-9, Tegmark, 2018, pp. 101, 122.
 Russell, 2019, p. 118; Fuchs, 2019, location 1712 points to something similar in Germany.
 2018, p. 319.
 Fuchs, 2019, location 1777.
 Yonck, 2020, location 2229.
 ibid., Location 2246.
 ibid., Location 2281.
 2018, p. 115.
 Yonck, location 2429.
 ibid., Location 2826.
 ibid., Location 2500.
 Zuboff, 2019, location 4503.
 ibid., Location 6043.
 Ferguson, 2017, location 221.
 ibid.., location 78.
 ibid., Location 420.
 ibid., Location 561.
 ibid., Location 157.
 ibid., Location 993.
 ibid., Location 2479.
 cf. Srnicek and Williams, 2016, location 1990.
 Ferguson, 2017, location 1110.
 ibid., location 1167.
 ibid., location 2014.
 ibid., location 2221.
 ibid., Location 2182.
 ibid., location 3069.
 Tegmark, 2018, p. 104.
 Susskind, 2018, p. 45.
 ibid. p. 130.
 ibid., p. 150.
 ibid., p. 228.
 ibid., p. 229.
 ibid., pp. 230-3.
 Foer, 2017, pp. 6, 149.
 ibid., p. 217.
 ibid., p. 75.
 ibid., p. 354.
 ibid., p. 154.
 ibid., pp. 195-7.
 Bastani, 2019, location 315.
 ibid., location 329.
 ibid., location 344.
 ibid., locations 523-8.
 ibid., location 757.
 ibid., location 775.
 Russell, 2019, p. 101.
 Fuchs, 2019, location is 636, 781.
 Bastani, 2019, location 1042.
 ibid., location 2305.
 ibid., location 2815.
 ibid., location 3178.
 Srnicek and Williams, 2016, location 227.
 ibid., location 403.
 ibid., location 1512.
 ibid., location 1579.
 ibid., location 2083.
 ibid., location 2182.
 ibid., location 2387.
 ibid., location 2866.