|James Hairston||Frank Levy||Christina Colclough||Stuart Elliott||Young Tae Kim||Mark Keese|
Last year Stephen Hawking controversially claimed that artificial intelligence (AI) will be “either the best or worst thing” for humanity. Indeed, while AI can help tackle global challenges and deliver considerable benefits, it also creates new challenges and questions, including for privacy, governance, research and work. To discuss these challenges and opportunities, as well as their policy implications, the OECD hosted the conference “AI: Intelligent machines, smart policies” in Paris on 26-27 October 2017, bringing together policymakers, representatives of civil society and AI experts from industry and academia.
One of the sessions focused on jobs and skills. AI matches or exceeds human performance in a growing number of domains, and several tasks traditionally performed by humans have already been taken over by robots and algorithms. Moreover, the general consensus is that the capabilities of AI will continue to grow and its use will become rapidly more widespread. The exponential growth of the capabilities and applicability of AI has raised concern about job automation and the possibility of massive technological unemployment, but also about its downwards impact on the wages of workers who are most at risk of being displaced. As framed by the moderator of the session, Mark Keese (Head of the Skills and Employability Division at the OECD): “Some commentators have raised the spectre of inventing ourselves out of existence by developing ever more powerful AI. But a more immediate concern is whether we are running the risk of inventing ourselves out of work.”
Stuart Elliot, from the US National Academies of Sciences, Engineering and Medicine, painted a particularly sobering picture. Working with a group of AI experts, he tried to estimate the extent to which current technologies can answer the literacy and numeracy questions of the OECD Survey of Adult Skills (PIAAC). Although he found that there are things which AI cannot yet do, he also stressed that many individuals cannot do them either. In fact, his research suggests that only 11% of adults are currently above the skill level that AI is close to reproducing.
Frank Levy, Rose Professor Emeritus at the Massachusetts Institute of Technology, added to these concerns. He argued that, with all the focus on what AI might do in the long-run, we risk are missing a lot of what is happening in the short-run. Indeed, those in lower- to mid-skilled occupations involving significant amounts of repetition have already been affected by technology and remain those at highest risk of losing their jobs. Across many countries, labour markets have been polarising, with a rise in the share of both high- and low-skilled jobs, on the one hand, but a fall in the share of medium-skilled, routine jobs, on the other (OECD, 2017).
Such upheaval in the labour market can give rise to severe political reactions and, argues Professor Levy, the adoption of AI and the policy response to it will largely depend on that. That nothing is inevitable in how and when AI is adopted was echoed by Christina Colclough, Senior Policy Advisor at UNI Global Union. She warned against “technological determinism” and urged the social partners to talk about and agree on the kind of future we want – she said “we shouldn’t go ‘belly up’ when we think about the future.”
Young Tae Kim, Secretary General of the International Transport Forum, tried to counter some of the pessimism and argued that “people have a tendency to focus on job destruction rather than on job creation”. While it is true that AI may take over some tasks and jobs, it also creates a whole host of new ones. Moreover, as James Hairston (Head of Public Policy, Oculus VR, Facebook) argued, AI is not only a replacing technology, but also a complementing one and, as populations age, it could help increase productivity.
Mr Hairston also stressed that making AI work for everyone requires having the right policies in place and, in particular, that governments needed to invest in people and skills so that individuals can seize grab the opportunities that lie ahead. He stressed that investing more in skills and facilitating greater labour mobility are were not new policy challenges. However, here again, Stuart Elliott sounded the alarm bell. While he agreed with Mr Hairston, he also stressed that “We do not have examples of education policies at scale that bring 80% or even 50% of adults above the current computer level.” If this is the case, then it either means that governments need to think about new tools and incentives for promoting adult skills or that skills policies should be combined with other interventions, including social protection and social dialogue, if AI is to lead to widespread improvements in well-being.