Visions of a technological utopia in which humans are free of the shackles of work and where purpose-built robots undertake a myriad of tasks in their place has long been a motif of science fiction. Yet the seemingly exponential advancement of artificial intelligence (AI) means that vision may soon become a reality. A report published last year by IPPR concluded that one in three jobs are thought to be at risk from automation within the next two decades – the general consensus amongst academics being that these will be primarily low-skilled jobs. Recently released research by the OECD puts that number lower, but still suggests a significant impact, concluding one in ten jobs are likely to be automated.
Whether the actual number of jobs to be automated is one in three, one in ten or something else, automation will necessitate some kind of shift in social policy, including welfare. Science fiction may have presented us with stark depictions of the process of automation, but it has little to offer about what that shift should look like – for example, how our welfare system will need to adapt to a labour economy in which AI plays a key role.
One rather radical policy solution which has been mooted is universal basic income. Sometimes referred to as a ‘zombie policy’ on the grounds that it simply will not die, the debate around automation has represented a new opportunity for those who support it to make their case. The policy itself is, in theory, a simple one: a welfare regime in which all citizens (or permanent residents) of a country receive a regular, livable and unconditional sum of money from the government. If automation takes jobs, so the argument goes, then people will need some kind of replacement pay and universal basic income will give everyone the security they need to survive. Support for UBI is found across the political spectrum – from perceived ‘radicals’ within the green movement and Silicon Valley, namely Mark Zuckerberg and Elon Musk (who has stated that “universal basic income is going to be necessary”), to centre-right politicos such as George P. Schulz and Henry Paulson. But of course, the obvious problem is the cost implications of providing every citizen with a basic income, necessitating either massive increases in income tax or, potentially more feasibly, using the revenue levied from the alternate policy discussed below as a source of funding.
Another solution which has been posited is a ‘robot tax fund’. This would involve taxing companies that profit from using robots and then using the revenue generated to fund either UBI or retraining for individuals who have lost (or stand to lose) their jobs to AI. The policy has received the patronage of Bill Gates as well as Jeremy Corbyn’s Labour Party. However, it is no less controversial than UBI – it received an overwhelmingly negative response when presented to the European Parliament in 2016, with the International Federation of Robotics fearing it would stunt innovation – “the idea to introduce a robot tax would have had a very negative impact on competitiveness and employment”.
Purported trials of UBI have been conducted but they are not truly universal, with only a handful of citizens being offered the basic income. Moreover, we do not yet have any significant data on the outcomes of these trials. Robot tax pilots are also limited. As a result, there is a large degree of uncertainty about how these policies would operate in reality – most crucially, what their impact on people’s behavior and choices around work would be. Arguably the largest obstacle in deciding the best way forward is that we simply do not have a grip on the issue – one in three jobs being automated is a very different problem to one in ten and the two scenarios would likely necessitate different policy responses. Until we know the true scale of the problem, it is challenging to talk meaningfully and evidentially about potential solutions.
Although the approach that governments ultimately choose to tackle this issue and the details thereof is naturally up for debate, what is not a point of contention is that a solution needs to be found. For all of the seemingly limitless benefits that AI integration into the labour economy has the potential to bring, it is crucial not to lose sight of the impacts it will inevitably have on society as a whole. In short, the current models of state welfare provision across the globe seem broadly irreconcilable with the rising assimilation of AI into the employment market and therefore reform of such models is necessary.