When you know a storm is coming, you have a sense of its likely form, speed, direction, potential impact, and possible duration. Armed with that knowledge you can then take in provisions, batten down the hatches and go to a place of safety. When that storm is in the form of artificial intelligence (AI) software powering smart or even superintelligent machines and robots that could impact every aspect of life on the planet, the challenge takes on a wholly different magnitude.
So, what’s really going on here and why are we talking about taxes for robots as a solution to the potential rise in long-term unemployment?
Here we explore ten key questions to help the reader get their head around the whole subject of taxing the robots.
What’s driving the debate?
The big issue here is the likely extent to which automation will reshape the industrial landscape, change the nature of work as we know it and drive up the number of people facing permanent unemployment.
On the one hand, we know smart technologies such as AI, blockchain, big data, cloud computing, hyperconnectivity, 3D/ 4D printing, smart materials, and synthetic biology are developing at an exponential rate. Individually and when combined, they will have an impact from automated warehouses and autonomous cars to computerised drug discovery and the diagnosis of Alzheimer’s disease ten years before the symptoms show – it’s already happening.
However, and crucially, we don’t know how far and how fast AI and these other disruptive technologies will spread. Furthermore, we don’t know how many jobs they will take out, we don’t know how society will respond (e.g. the Uber backlash), we don’t know the extent to which firms will retain people when they automate, we don’t know how fast the new sectors will grow, and we don’t know how many new jobs they will create. In practice we are pretty clueless.
So why act now?
What we do know is that jobs are already going, more will follow and it would be a reckless government that didn’t think about how to address the challenges.
We already know that to compete in the emerging global economy, we need to change the nature and focus of education at all levels and prepare adults for roles in the new sectors – which will mainly be higher skilled as the robots will most likely do the rest. So, it’s reasonable to at least explore the scenario of rising technological unemployment over the next decade.
Realistically, this means we’d need to fund either a higher total unemployment benefit bill or the provision of some form of guaranteed basic income and / or guaranteed basic services. In this scenario, fewer people working means they are likely to be paying less overall income tax which means we have to fund the revenue shortfall somehow – that’s assuming we want to maintain the current level of public service provision whilst also covering the higher unemployment costs.
How is the UK government responding?
Britain’s ruling Conservative Party is loath to acknowledge the possibility of rising unemployment due to automation. The hope is that encouragement of free markets and lower corporate tax rates will drive business growth and employment. They believe that unemployment costs will be met through revenues from corporate and individual taxes coupled with VAT.
In contrast, the rising number of young members in the opposition Labour Party are concerned about the impact on their future – spurred on by already high levels of youth and graduate unemployment. They are keen to ensure Britain doesn’t go into the kind of decline we saw with Greece and Spain.
In response, and acknowledging the fundamental changes taking place in the industrial economy, Labour has been mooting the idea of “robot taxes” to finance the cost of adult retraining, education transformation and unemployment provisions. The argument is that robots should be taxed because they will be considered as something that creates value for the owner, like property, and if firms are cutting headcounts, then they are likely to be making higher profits. Furthermore, the belief is that those who will receive the benefits will spend that money with the firms who paid the robot taxes.
What would robot taxes pay for?
Clearly, the primary purpose should be to address the societal consequences of job automation. So, the most obvious application would be to fund unemployment benefits or guaranteed incomes and services.
Alongside unemployment costs, there is a strong argument that a significant proportion of the revenue from robot taxes should be channelled directly into public education. This would create a positive role for robots in society, which would be to pay for public schools and universities.
A robots tax could help pay for a new approach to education which develops the whole person, not just the ‘future worker’. These would include life skills (cooking, health and household management), interpersonal skills (listening, leadership, writing) and self-awareness (mindfulness, meditation, mental health strategies). The underlying principle is that we should use the value of automation to benefit society and prevent future problems.
What is the likelihood of governments around the world introducing a robot tax?
Some governments have already started to think about the spending side of the equation – the human consequences of automation – exploring everything from new approaches to adult education to encouraging the creation of start-ups.
Canada, Finland and Germany have also been experimenting with different forms of guaranteed or universal basic income (UBI). These are relatively small experiments – the intention is to learn about them before they are required. The experiments are looking at different funding models, whether any access conditions should be applied, and the impact on mental health, domestic violence, crime, and community cohesion. Such experimentation seems eminently sensible as an input to any nation’s debate on the topic.
At a broader tax policy level, across the world, rapid automation must be seen as one very important driver of change to nations’ tax collection regimes. Clearly the public spending policy decisions of these governments will also have an impact. Hence it becomes critical to explore different possible scenarios to understand the likely spending requirements and revenues under a range of different conditions. Governments can then examine both their spending priorities and possible revenue instruments.
Who might lead the way and when might it happen?
By 2030 the possible pace of change means these taxes could well be commonplace in many industrial nations.
Countries that are embracing automation and the digital era in all its forms such as South Korea, Japan and Singapore might be among the first to implement some form of automation taxation mechanism. China is saying little right now, but it has the capacity to enact policy rapidly should the need arise. Whereas, the overt and hidden political power of the Indian super-corporates means it would be a very late adopter.
In Europe, nations such as Estonia, Finland, Sweden, Denmark, Iceland, and Germany are likely to be among the first to revamp their tax systems in this way.
Whilst many in Silicon Valley argue in favour of taxes for robots, the US is likely to face strong resistance to such changes. Indeed, it could well be among the last to go down this route and might conceivably not do so at all without a fundamental change in its governance and electoral systems.
How might such taxes work in practice?
The going in point here should be to evolve a more flexible approach to creating income to fund future public services. The basis of corporate taxation could become even more complex with systems applying AI to large multi-variable data sets to establish a tax liability based on the sector, revenues / profits per employee, the number of people employed, and geographic location. The algorithms could also take account factors such as expenditure on training and retraining current and former employees, the support given by firms to start-ups, the level of employment created further down the value chain, and the amount of tax paid by the firm’s employees.
Perhaps evaluation of a business’s broader impact on society could also factor into the level of taxation applied to its profits – such as the actual level of human employment, local and national social responsibility, environmental impact – so that tax paid is based on the outcomes of a business’s operation across a range of different domains.
Some measure of net added value could also be considered. For example, a firm may train its employees so well that they go on to higher paying jobs elsewhere or to generate employment and tax revenues by starting their own business. How might their taxation be assessed relative to a firm who invests little in people development and whose staff cannot find jobs elsewhere when made redundant?
An interesting scenario to explore would be the possibility that AI could create the opportunity for governments to recover public spending commitments pro-rata from every tax payer and corporation in the country – purely based on individual incomes or business revenues.
The key here is modelling a variety of different approaches to see which produces the fairest and most transparent system.
There are also precedents we can learn from. For example, the British pharmaceutical industry has paid a levy based on revenue or capital employed on its supplies to the NHS with a series of allowable and dis-allowable expenses. This approach has been designed as a mechanism to control profits on medicine supplies to the NHS (while seeking to reward investment in R&D) and a similar approach could be taken with companies and the level of automation they employ, versus their “investment” in people.
What potential risks and drawbacks are there?
It is already being cast as unbridled socialism, communism or Marxism by many proponents of low taxes and free markets. However, at present, no viable alternatives are being put on the table.
At the operational level, it could be costly and complex to implement and opponents will look for any shortcomings to cast it off as a failure. The prevailing corporate mind-set is often to base multinational operations in lower tax markets, so competition for the hosting of multinational organisations could intensify without global agreements. Inevitably, many will look for ways to minimise their tax payments and a range of advisory services and schemes will spring up to help firms do so.
Failure to implement a viable system or a workable alternative could have disastrous consequences for governments, leading to potential reductions in public service provision and even the failure of some economies.
What are the potential benefits?
A solution will be required if unemployment does rise and therefore government revenues decline. Whilst robots taxes may not be the ultimate answer, it is the only clear policy idea that is even being mooted today for what is an increasingly pressing societal issue.
Ultimately, the notion of taxation based on automation could prove to be a catalyst for more socially responsible “carrot and stick” approaches to corporate tax. Maybe the application of increasingly sophisticated AI could be the critical enabling technology to providing a fair and transparent system with no potential for avoidance or manipulation by individual firms. Indeed, AI could one day give us even smarter tax systems that none of us can even imagine today. Perhaps the fully automated corporation or Decentralised Autonomous Organisation (DAO) of the future may see its prime directive to serve humanity as a whole and maximise its contribution to society.
How do we get started?
The first stage must be to run some serious computer simulations of different scenarios for the pace of automation and the impacts on employment in different countries around the world. These could be used as input to the development of economic models to explore the funding requirements of different public service strategies and how they might be met. If there’s a shortfall between what’s required and what could be collected under the current taxation regime, then the potential for different robots tax models (and any alternatives) could be evaluated and the likely implications assessed.
Artificial intelligence is creating the tools that are driving the pace of automation and the prospect of increased unemployment. Equally, AI tools could also be used to design and develop new approaches to taxation that could help us address the societal consequences of technological disruption and ensure a very human future for all.
Rohit Talwar, Steve Wells and Alexandra Whittington, Fast Future