Benjamin Joyeux: Tiffany Blandin, you have just published Un monde sans travail?  (A world without work?), in which you show the way that many companies are preparing to get rid of a large proportion of their employees thanks to massive advances in robotics and artificial intelligence. Indeed, as this new ‘revolution’ entrusts various tasks to non-human entities, it may well signal the end of work and wage-earning as it we have known it for the last two centuries, rendering thousands of workers useless and plunging them into unemployment and a precarious situation. Firstly, according to your research, how much has global automation progressed so far and is there a real danger of “the end of work”?

Tiffany Blandin: We have been hearing for decades about robots that are supposed to replace workers for manual tasks in industry. What is new is that precisely the same thing will happen, but for the intellectual tasks of office employees.

Indeed, since the early 2010s there have been amazing advances in ‘deep learning’, a form of artificial intelligence – AI – based on neural networks. This has made it possible to create programmes that perform extremely well for tasks that involve retrieving information from a significant quantity of data (big data). These programmes are highly effective for anything to do with voice or photograph recognition, for example.

This means that a large proportion of bureaucratic tasks carried out by employees in the workplace can now be automated: setting up appointments, taking minutes, database analysis, writing simple texts, customer relations, and so on. The general idea is that overall considerably less labour will be needed if employees save time on such tasks.

In view of the promise of these new technologies, big web companies have invested an enormous amount of money in them. They have built AI laboratories, bought hundreds of start-ups in the field etc. Indeed, the Internet giants are marketing what are known as the “building blocks” of AI, that is to say lines of code that any developer can then put into a programme. In fact, start-ups around the world are coming up with all sorts of highly specialised smart software, which they then hope to be able to sell to companies for a high price. For example, the company Inbenta offers automated customer service, which is perfect for banks or insurance companies. Deepomatic sells a smart camera, which could replace jobs in surveillance and observation that are still performed by human beings. All these programmes are currently being marketed, and they are in a test phase at large companies.

The striking thing today is that all economic sectors are involved. Studies estimate that between 9% and 47% of jobs could be automated in the short term. The higher figure was the result of research by the strategic consultants McKinsey. This has led French researchers such as Bruno Tebopul and Raphaël Liogier to conclude that “creative destruction” (the idea that eliminated jobs will be replaced by new ones) will not work in this case. It is clear from the ‘true’ unemployment figures (the total number of people registered at the Pôle Emploi – the French job centre) that this theory, devised by the economist Joseph Schumpeter, no longer applies: in France we have gone from 4.2 million unemployed people 20 years ago to more than 6 million today.

Clearly, many new jobs have been created. However in reality, the vast majority of these new jobs do not offer full employee status with all the social protection that it affords. A study has shown that in the United States, between 2005 and 2015, 94% of the ‘jobs’ created were actually alternative forms of employment, particularly freelance work and online platforms. So the question arises of the quantity – but above all the quality – of the new jobs created, compared to those that are disappearing. There is also a fundamental question: what will happen when there are many more people out of work than there is work available? As there is no minimum wage for freelance work, and everything is negotiated between two unequal parties, the law of supply and demand applies. This means wages could collapse, and this is already the case in some sectors, such as content provision. You only have to go to some platforms such as redacteur.com to see it for yourself. Dozens of candidates put themselves forward for underpaid assignments there. In the future, this imbalance could apply to most sectors.

Now that the question is starting to come up in public debate (for example Elon Musk and other tech leaders recently urged the UN to act upon the perils represented by so-called “killer robots”), having previously been the preserve of science-fiction films and novels, do you think it is being tackled in the right way? Have we already missed the boat on this wave of technology?

Indeed, the term ‘artificial intelligence’ fires up people’s imaginations, and this prevents them from seeing the most urgent questions. Science-fiction films and futuristic series and novels immediately conjure up ethical questions like: should we let machines kill or judge a human? In the future will we be able to ‘augment’ intelligence by implanting chips in people’s brains? Or: will machines take power one day?

Last year, many big newspapers featured the story that two Google AIs had developed their own language to the extent that it could not be deciphered by a human being. In fact, the researchers at the company were quite capable of doing so. Examples like this are widely reported every week.

Recently I was invited to the set of French television programme Arrêt sur images, which was doing a programme on AI. There too, no sooner had the question of the effect on employment been mentioned, than the chief editor preferred to focus on more ‘sexy’ aspects of the relationship between people and machines. “Employment? People know about that already”, he said to me after filming.

I do not agree. A large proportion of the population does not know what is happening – not least because politicians don’t express it clearly. In April 2017 during the debate between the five main French presidential candidates, all of them – except socialist and Green candidate Benoit Hamon – promised to restore full employment. What is more, when public figures admit that millions of jobs could be under threat, as Emmanuel Macron did in the end, they always add that new professions will also appear – still with an unshakeable belief in ‘creative destruction’.

The problem is that nobody knows how many new jobs will be created. On this subject it is very instructive to look at the studies on the professions of tomorrow that are regularly published. Some have been prepared by PR agencies that are short of publicity, and they discuss absurd professions such as ‘productivity consultant’ or ‘urban shepherd’.  These studies get masses of shares and likes on social networks, which shows that people want to believe at any cost that technology cannot threaten the balance in our society.

Although all the questions above should indeed be posed at some point, the most urgent issue is that of the economic and social consequences of the development of robotics and AI, because it is now that everything hangs in the balance. Many international figures have already raised the alarm: Andrew McAfee and Guy Standing to name but two. Silicon Valley bosses Mark Zuckerberg and Elon Musk have expressed their support for a basic income, a proposal that is closely linked to the future shortage of jobs. One could also cite Jim Yong Kim, the President of the World Bank, who maintains that “two thirds of jobs could be eliminated because of automation in developing countries.”

Isn’t the main problem with this question the issue of “democracy”, as we plunge into automation controlled by experts and large companies rather than citizens, a bit like the nuclear question?

Governments are investing massively in the development of AI, in the United States, Canada, China, South Korea, and maybe soon France and the rest of Europe. A French national consultation called France IA, intended to create a ‘French model’ of AI, was launched at the start of 2017. Its recommendations? To create grants for researchers and to build a European artificial intelligence laboratory endowed with 100 million euros. How can one explain to citizens and tax-payers that public money is invested in AI, while it is precisely this technology that will threaten their jobs? There is still a lot of inconsistency and we are very schizophrenic.

Furthermore, a France IA working group was intended to examine the economic and social consequences of the democratisation of AI. The members of the working group used the lower end of estimates from existing studies (9%). The effects on employment would therefore be minor – even if this figure represents 2.3 million jobs in France. These conclusions are not surprising. The working group was mostly made up of people from AI start-ups, so they had no interest in alerting the public.

To summarise, political leaders have some concerns: “the world has started moving towards artificial intelligence, we must not get left behind”; start-ups have other concerns: “an enormous market is opening up.  There are good opportunities to take.”; and finally, their client companies: “from now on we have to implement smart solutions, or else the competition will do it and we will disappear.”

In all of this, nobody is currently representing the interests of a typical wage-earning citizen. It is all the more problematic as the question is very complex (you need to understand what AI is, the extent to which it can replace human beings, if companies are ready to adopt it etc), so it is difficult for the public to come to a free and informed opinion. Unfortunately, the loudest voices are still those of political and economic leaders who intentionally play down the danger.

How would you envisage appropriate rules, that are both on the right scale (the robotic revolution that is underway is being carried out on a global scale by transnational actors), and also sufficiently fluid to adapt to an extremely fast-moving phenomenon? For example, we can now see how legislators almost everywhere are finding it difficult to follow the changes caused by advances in the Internet.

In France, the latest reforms to the labour code have helped further unravel the status of employees. And trade unions have not been able to contain the movement. Economists such as the late Bernard Maris have shown that this is largely because it has been impossible to reduce the unemployment rate over the last 30 years. It is therefore hard to imagine how workers will be able to exert pressure to oppose vast social plans or to demand compensation for the shifting of revenue from labour to capital.

One might think that the safeguards to limit the damage can only be set up through politics. However, as I was told by Malik Ghallab, an AI researcher at LAAS-CNRS in Toulouse, AI is progressing exponentially, while society and politics move at a much slower pace. He maintains it is therefore all the more urgent to legislate now, before massive unemployment and maladministration cause too much damage in society.

But we live in a globalised economy, which limits what is possible. Thus the proposal of a ‘robot tax’, which would be imposed on companies that automate jobs, is not taken at all seriously by many politicians, as it would curb the competitiveness of companies in the country where it was implemented.

Currently there is only one political response to the crisis: an overhaul of the system for initial and continuous training. It is a question of training young people in IT, and enabling unemployed people to gain digital skills. This is based entirely on the myth of destructive creation. The first problem is that it strengthens social, generational, and geographical inequalities. How can you tell a 45-year-old order picker, like those I met in Lectoure in the Gers region of France, that they have to learn to code and move to Paris near the start-up incubators? The second problem, which has been raised by researchers, is that there is no point giving digital training to the entire population if there are no jobs at the end of it.

Some politicians, both male and female, are starting to explore options. There is Benoit Hamon, a socialist candidate who clearly raised the debate during the last French presidential campaign. In the European Parliament, the Luxembourg MEP Mady Delvaux published a report on the subject, proposing a resolution to debate the robot tax and universal income, but it was rejected by the Right.

An unconditional basic income currently appears to be the most serious initial solution to the loss of revenue by a significant part of the population. However this type of revenue may not be the perfect response, as it could create a very non-egalitarian society, with shareholders and technological workers on one side earning a very good living, and the rest of the population on the other, receiving a minimal amount so they continue to consume and do not revolt too much.

Some sociologists such as Dominique Méda have also highlighted the fact that work has a primordial social function as well, as a form of development and self-realisation. This means it is difficult to deprive a large part of the population of it.