It’s not about technology, it’s about us!

An article from Linda Kool and Rinie van Est about working in the robot society

Will robots take over work from us or enrich it? What do we want the robot society to look like? This article considers various scenarios for the future that show there’s more than just the nightmare and the ideal.

Let’s see…. According to this BBC online tool (based on studies by the Oxford Martin School and Deloitte), the work of researchers still continues to be difficult to automate. The likelihood of our work at the Rathenau Institute being taken over by a robot in the next twenty years is only seven percent. But if you’re a payroll manager or bank clerk, the risk of your job soon being taken over by a smart machine is 97 percent.

Microelectronics will cause major shifts in the labour market and change the nature and content of a lot of work

People have always been worried about losing their job because of technological advances, for example way back at the start of the Industrial Revolution. More recently, the Dutch government set up an Advisory Group in 1979 chaired by the man who gave his name to our Institute, Prof. G.W. Rathenau, because it was concerned about the socioeconomic impact of microelectronics, in other words ICT.

The Advisory Group concluded that “microelectronics will cause major shifts in the labour market and will have a powerful impact on the nature and content of a lot of work.” To cope with the expected “significant” structural unemployment, employees would need extra training, retraining or further training. There also needed to be a vigorous innovation policy and a meaningful division of labour.

Recent decades would seem to have confirmed the correctness of the Advisory Group’s conclusions: some jobs have disappeared but others have been created. Indeed, one can say that in general the use of microelectronics has contributed greatly to increased prosperity, as is shown, for example by the Rathenau Institute’s report Working on the robot society. The Institute was commissioned by the Dutch House of Representatives to investigate the impact of robotisation and automation on people’s work. The report surveys the main scientific research insights so as to facilitate well-informed and topical debate on this subject. It forms part of the “Smart Society” theme within the Institute’s 2015–2016 work programme.

The Dutch Government responded to the report in December 2015. One of its comments was: “In the long term, the changes may in fact be greater than currently envisaged, and may occur in other areas.”

The second machine age is characterised by machines that also replace brainpower

The big question is whether the old belief that technology quickly creates new jobs will remain true. In their book The Second Machine Age, Brynjolfsson and McAfee (2014) argue that that will probably not be the case.

The first machine age, beginning in about 1800, was characterised by machines that could replace muscle power, enabling them to take over hard and repetitive physical work from humans. The second machine age, which began in about 1980, is characterised by machines that can also replace brainpower, meaning that they can take over brainwork from people.


Initially, this involved simple administrative tasks (see, for example, the report by Statistics Netherlands with figures on the Dutch situation), but ultimately more complex tasks such as writing reports, inventing cooking recipes, or diagnosing diseases will also be automated. This means that the impending wave of automation will flood over us all, regardless of our education or sector.

New organisations feature different employment relations

Using smart machines and other technologies also leads to different ways of organising work, for example algorithms to process “big data” and platforms such as Uber and Airbnb that can coordinate supply and demand within the cloud. Customers can already do more for themselves, such as scan purchases and bank online. The advent of the block chain (the technology behind the bitcoin) may eventually mean that “trusted third parties” will no longer be needed for carrying out transactions, so bankers, notaries, accountants, and tax inspectors will also need to look for a different job.


Another effect is the emergence of organisations with few permanent staff and little in the way of capital assets – including Uber, Airbnb, Helpling (home cleaning), and meal-sharing services – all of which use smart software to coordinate supply and demand. This is leading to a boom in the number of freelancers, a boom that had already started in such traditional sectors as building, postal deliveries, and journalism.

These new organisations feature different employment relations to in the past. People are no longer actual employees but are available on demand and often bring their own equipment with them. “Individual contracting” leads to heated discussions about working conditions, quality of work, compensation, working hours, social security and pensions, and about arrangements regarding refresher training and privacy. And judging by statements by Presidential candidate Hillary Clinton, this issue even plays a role in the US election campaign.

A society with robots as indispensable helpmates, as liberators, and as lifesavers

Books like The Second Machine Age have caused quite a stir in the Netherlands. In May 2015, a number of people including the eminent cosmologist Stephen Hawking and Internet entrepreneur Elon Musk (Tesla) published an open letter in which they implicitly warn of the risk of smart machines outstripping humans.

But are things are really as black and white as they may seem? Robots stealing jobs and machines taking over the world (like the VIKI supercomputer in the film I Robot) present a very one-sided and rather depressing picture of current developments in ICT and robotics. It’s as if there’s only one possible outcome of those technological developments, namely Modern Times 2.0 , with a contemporary Charlie Chaplin as the slave of the machine (see this recent speech by the Dutch Minister of Social Affairs and Employment, Lodewijk Asscher).


But Mr Asscher says that things can be different. We needn’t look forward to a society in which we are the collective slaves of technology but to one in which robots are indispensable helpmates, liberators and lifesavers, and in which human insights remain irreplaceable. The present article links up with that conviction and shows that a whole range of future scenarios are possible. It presents the options that we can choose from. But we must first actually want to make a choice and secondly work hard to achieve the desired aim.

Big data means smart machines can train themselves and get better

But before we look at the necessity to choose and the need for some hard work, we should first say something about the various technologies and especially the cross-fertilisation between them. As we’ve seen, it’s not just a matter of developments in robotics and artificial intelligence but also the collection and automatic interpretation of big data, the unlimited computing capacity in the cloud, and the rise of Internet platforms.

Thanks to big data, for example, smart machines have enormous quantities of data at their disposal that enable them to train themselves and to get even better. DeepFace, for example, utilises the enormous number of photos on Facebook so as to recognise faces increasingly accurately – indeed, it’s now as accurate as the average human.

Algorithms like this will soon enable robots not just to recognise faces but also to learn to interpret situations and to anticipate possible changes. Of course, “unintelligent” robots can already be used effectively in production, but the process needs to be adapted to the various possibilities and especially to the limitations of the robots.

Software developers share their algorithms, data, and hardware designs via online platforms

Thanks to the cloud, robots can expand their memory and computing capacity for recognition tasks (What’s this?) or for movement planning (Where do I need to go?). Google’s robotic car, for example, is in constant contact with servers so as to find its way using online maps and satellite data.


Robots can also collaborate and share knowledge via the cloud. In Amazon’s warehouses, for example, the robots “consult” one another to decide which of them will transport a particular item, so as to avoid running into other robots.

Online platforms enable engineers and software developers to share their algorithms, data, and hardware designs. One example of this is the Robot Operating System (ROS), a platform for developing software for robots. In recent years this has become a standard, greatly accelerating the development of robot software.

According to Martin Ford, author of The Rise of the Robots (2015), “the migration of leading-edge artificial intelligence capability into the cloud … is almost certain to be a powerful driver of white-collar automation.” And as if to confirm that, Google recently made the source code for self-learning software publicly available in the cloud.

Collaboration needn’t be restricted to physical work

Improvements in artificial intelligence and robotics (dexterity) enable the development of collaborative robots (co-robots) that allows robots to step out of their industrial ‘cage’. Such co-robots will not need to be shielded from their human colleagues but can work side by side with them. Those same human colleagues will be able to teach the robots how to do things without needing to programme them, simply by demonstrating the actions that are required.


So collaboration between humans and robots needn’t be restricted to physical work. As we have seen, the second machine age is characterised by the phenomenon of smart machines that can also perform brainwork tasks, but that doesn’t necessarily mean that they will threaten the livelihood of brainworkers.

A financial adviser, for example, can leave the task of finding the best investment opportunity to a smart machine while himself concentrating on a sound decision for his clients. Moreover, a human – for the moment, at least – is often better able to assess whether someone is creditworthy. After all, a computer initially refused an application by Ben Bernanke (ex‑chairman of the Federal Reserve) to refinance his mortgage.


So like co-robots that can take over boring, exhausting, or dangerous physical work, smart machines can enrich the work of brainworkers. As a “chef”, Watson – the successor to IBM’s Deep Blue chess-playing computer – produces interesting recipes based on available ingredients, but it did receive “prompts” from a number of leading chefs.

Active policymaking and investment are needed, certainly in the transitional period

We can summarise by saying that there are a number of technological developments in robotics, artificial intelligence, and related fields that will affect the way we work (both physical work and brainwork) and the way the work is organised. In public discussion, those developments lead to anxious faces and sombre predictions for the future. But are those concerns justified?

Well,… yes and no. Yes, because groups of people may lose their jobs. How many that will be is uncertain, while retraining demands effort and can lead to a reduction in earnings. So active policymaking will be necessary as regards retraining and extra training, certainly in the transitional period. Similarly, investment will be necessary in primary and secondary education to ensure that the knowledge and skills of people entering the labour market are in line with changing demand.

But also… no, because other scenarios are possible besides just the worst case scenario of rising unemployment, a vanishing middle class, and increasing alienation.

At the heart is the “business as usual” scenario familiar from history

Based on the various technological possibilities, we have identified a number of scenarios for the future (see the blue and red cloverleaf figure showing the range of scenarios). At the heart of the figure is the “business as usual” scenario already familiar to us from history. Past experience has shown that new technology first destroys jobs but quickly creates new ones. But “business as usual” doesn’t mean that no policies and action will be needed to ensure that the transition is as favourable as possible for everyone.

In addition to this “business as usual” scenario, we have also sketched a negative and a positive scenario for the future. In doing so, we have focussed mainly on the interaction between humans and smart machines and the role of Internet platforms in the distribution and organisation of work.

Amazon’s robotised warehouses: an example of far‑reaching systematism

Far-reaching rationalisation and automation in the workplace may mean that people will only be necessary to run the smart machines, with the machines becoming tyrants that not only prescribe what has to be done and how, but that also constantly monitor the people, just like they do the smart machines themselves.

An example of such far-reaching systematism (see the blue and red cloverleaf) are Amazon’s highly robotised warehouses, which the undercover journalist Jean Baptiste Malet says form a “univers chronométrique”.


Another example is the modern scheduling software that uses big data virtually real time to schedule the workforce. In practice that led to extremely variable working hours, sometimes communicated to employees only at the last minute. The New York Times reported in 2014 about the dire consequences that constantly changing working hours – announced only at the last minute – can have on an employee. Fortunately, consideration is now being given to how the system can also take account of the employees’ preferences.

Marc Andreessen – one of the founders of Netscape, the forerunner to Internet browsers such as Google Chrome, Firefox, and Internet Explorer – foresees a dichotomy in the workplace between highly educated people whose work is enriched by robotisation and automation and people with secondary or lower education who have to do the tedious and mind-numbing work that machines can’t (yet) do.

That dichotomy in the workplace can be seen as a (further) division of society into highly educated people with a good, “enriched” job and those with secondary and lower education who have to make ends meet by taking on several low-paid little jobs.

Martin Ford (2015) takes this a step further, asserting that we are moving towards an era of techno-feudalism (see the cloverleaf figure) – a scenario in which a small elite of very rich developers and technology producers take all the decisions. It’s what The Economist referred to in 2015 as the “Robber Barons and Silicon Sultans“.

Digital exploitation creates a new underclass that lags behind in all ways

Platform technology can further reinforce that dichotomy. The Internet has made it easier for people to find flexible work, not only locally but worldwide. This can lead to a situation in which they compete globally for little scraps of work for less and less pay, a kind of “digital sweatshop” (Zittran 2009) or Mechanical Turk economy (Reich 2015) in which people carry out mind-numbing mini-tasks at random times for just a few cents.


In this scenario, digital exploitation creates a new underclass that lags behind in all ways. That class is sometimes referred to as the “precariat” a portmanteau word combining “precarious” and “proletariat”. The precariat is susceptible to populism and extremism due to short-term jobs, low incomes and little social security, combined with the lack of a political voice.

Apart from creating social isolation (due to the addictiveness of competitively performing subtasks), the Mechanical Turk economy also lacks any overview of the bigger picture or any solidarity with it. People become alienated from their work and might not even know what they are producing.

This is also evident from the fact that Amazon’s Mechanical Turk service for coordinating the supply and demand for routine chores is also being used to produce spam or endorsements intended to make a political movement seem bigger than it really is. The people carrying out the work don’t actually know what they are doing.

Technology can also be used for a “paid spare time” scenario

The combination of smart machines and platform technologies can therefore lead to systematism, digital exploitation, alienation, and a society that can be described as techno-feudalism. But the self-same combination of technologies can also produce an entirely different and more positive scenario for the future.

The doomsday scenario of systematism can be diverted, for example, towards job enrichment, with humans and machines complementing one another as partners. Examples might be a car mechanic using a computer to diagnose problems, or smart search engines to help lawyers and notaries to quickly look up relevant jurisprudence.

Instead of a further dichotomy in which a denim-clad elite of feudal “silicon sultans” take all the decisions, the technology can also be used to achieve a “paid free time” scenario. But that will require considering new socio-economic policies, for example the introduction of a basic income or a negative income tax (as argued by Martin Ford, Erik Brynjolfsson and Andrew McAfee). It might also involve giving people a greater share in the growth of labour productivity (see, for example, this article by Prof. Richard B. Freeman of Harvard University). With measures such as these, a working week of ten or twelve hours can assure people of an income that is sufficient to live well on.

A scenario with people as micro-entrepreneurs with a global sales market

Platform technology offers people the opportunity to be their own boss and to work at times they choose, often at a location of their choosing. Linda Gratton (London Business School) refers in this connection to the “micro-entrepreneur revolution”, a scenario in which people are micro-entrepreneurs with a sales market that is global, i.e. not just in the western world but also in developing countries.


Platform technology also offers opportunities for increased specialisation. The Topcoder platform, for example, splits up a programming assignment into a number of tasks that can be presented on the platform as a puzzle. Enthusiastic professionals can then enrol to solve the puzzle in competition with others. There are attractive cash prizes to be won.

Such platforms can enable people anywhere in the world to discover or develop their skills and make money with them. Topcoder is also an “incubator” for talented programmers – one where large companies’ recruiters like to scout around. That opens up opportunities for people who usually have less chance of finding a job, such as young people, the elderly, and members of minorities.

Future scenarios show that we can guide technological developments

Based on these technological developments, one can envisage both a utopia and a dystopia. In both cases, it seems as if these developments are simply steamrollering us uncontrollably. By sketching the various scenarios for the future, we want to show that technological development is not an uncontrollable steamroller but that we can in fact guide the various developments: we can make choices.

In the Netherlands, we see that already happening. Varies parties are now actively thinking and preparing for a ‘robot society’.  Major cities, for example, are thinking about “smart cities” and about how they should deal with initiatives like Uber and Airbnb. The Dutch Association of Insurers has set up an insuranceLab so as to experiment with apps and big data, while network manager Liander now has a Livelab to develop smart grids and started a project to ask their technicians what they’d like to see robotised in their work (what is dull or dangerous work?), and subsequently try to design robots based on their employees’ input.

Three key requirements: encourage innovation, ensure training, and reflect on wealth distribution

So what now? We believe that policymakers and politicians need to look beyond just the horror and hype scenarios that all too often dominate the media. We want to challenge society to engage in a fundamental discussion about the meaning of work in the robot society. There are three key requirements: we need to stimulate innovation, invest in training and education and reflect on wealth distribution (see the Rathenau Institute’s report Working on the robot society).

Those conclusions are basically hardly any different to those of the Rathenau Advisory Group in 1979 about the advent of microelectronics. The point is that we as a society can decide for ourselves on how we want to use technology. As Melanie Peters (director of the Rathenau Instituut) puts it: “What society and our lives should look like is up to us. It’s not about the robot. It’s not about technology. It’s about society.”

This article builds on Rathenau Institute’s report Working on the robot society. A longer version of this essay can be found as the chapter on “Opportunities and threats: nine perspectives on working in the robot society” in the WRR Investigation “Mastering the Robot” (in Dutch).

To see more of what robots can do, here are clips of them checking the stock in a supermarket, preparing hamburgers, and harvesting cucumbers.