The recent OECD Forum virtual event: How AI Might Change Our Jobs & What We Can Do About It explored how we could harness the many benefits of rapidly developing AI while also addressing its significant risks. It is an important moment to discuss this as the adoption of AI in the labour market still remains relatively low, but costs are falling and more and more workers have AI skills, which leads to the expectation that OECD economies could be on the brink of an AI revolution. This makes, as the OECD Employment Outlook 2023 emphasises, the need for decisive action to develop policies to reap the benefits AI can bring to the workplace while addressing risks for workers’ fundamental rights and well-being urgent.

Benefits and risks

Speaking directly to stakeholders in the manufacturing and finance sectors about their experience as early adopters of AI, the OECD found opposing views on the benefits and risks. As AI adoption is still in its early stages, many companies choose to adjust the workforce through slowed hiring, voluntary quitting and retirement rather than through direct dismissal. Some companies even believe that AI could alleviate problems associated with an ageing population and labour shortages.

Nonetheless, the potential for worker replacement remains high. As a result, three out of five workers in the manufacturing and finance sectors in seven OECD countries are concerned about losing their jobs to AI in the next ten years. Separately, the OECD estimates that approximately 27% of total employment comprises jobs at high risk of automation.

Despite the widespread anxiety about the future, Mark Keese, Head of the OECD Skills & Employability Division, who oversees the OECD’s Future of Work initiative, emphasised AI’s potential for a positive impact on job quality. It’s not just about more or fewer jobs – it’s also about AI creating new jobs and ensuring the quality of these is up to standard. Indeed, nearly two-thirds (63%) of workers interviewed by the OECD reported that AI had improved their enjoyment at work. As menial tasks, from data entry to scheduling appointments, writing minutes to analysing legal documents and anything in between, are eliminated, workers can focus on more complex and interesting aspects of their jobs.

On the other hand, when tedious tasks disappear, workers can be left with more intense and high-paced responsibilities, leading to an increase in stress and pressure. What’s more, the use of algorithmic management (AI managing workers) comes with serious ethical challenges around data protection, transparency and explainability, bias and discrimination, automatic decision making and accountability. In some cases, AI hiring practices could be discriminatory, with anecdotes of built-in biases against women, minorities, and people with disabilities.

Rise of specialised AI skills

As AI has increased the importance of specialised AI skills - with some employers considering the lack of these skills a barrier to the adoption of AI - investment in education and training will be critical during the transition. OECD research shows that outcomes are better where workers have been trained to interact with AI, and where the adoption of technologies is discussed with them. Along with providing training, urgent policy action is needed to address the risks AI can pose to workers’ privacy, safety, fairness, and labour rights. The OECD AI Principles for responsible stewardship of trustworthy AI, adopted in May 2019 by OECD member countries, can guide efforts in terms of regulation and guidance.

As policy experts are concerned about the lack of specificity and enforceability of policies regarding AI, it will be crucial to cooperate internationally to avoid fragmentation of efforts and potential harm to innovation and regulatory gaps.

AI should never be used to check if you are angry, happy or unhappy, or stressed, as this type of surveillance would be infringing core fundamental rights, privacy as well as human dignity

AI surveillance

Isabelle Schömann, Deputy General Secretary at ETUC, stressed that AI is a product placed on the market, meaning it should respect legal standards and the rights of the people to whom it will be applied. She highlighted the importance of social dialogue, and collective bargaining to reach agreement between workers and employers on what protections are needed. She was concerned that while AI may not yet be so widespread in the workplace, AI is already being used to hire and fire workers, and in surveillance of workers in warehouses. Another dimension of concern is that AI is predictive, and might be used to predict whether a worker wants to unionise, with potential negative consequences in terms of employment contracts.

Brando Benifei, Member of the European Parliament and co-rapporteur for the EU Artificial Intelligence (AI) Act, mentioned that for him it was also key that employees be consulted whenever an AI system is adopted in their place of work. He highlighted that MEPs want to ban the use of AI-enabled emotional recognition at work, in schools and universities. AI should never be used to check if you are angry, happy or unhappy, or stressed, as this type of surveillance would be infringing core fundamental rights, privacy as well as human dignity.

While young people want to learn about technology, they are concerned about how potential future employers will use AI to control people rather than to empower them

Inclusive employment

Soumitra Dutta, Dean & Professor at Said School of Business, mentioned that working with young people, he sees a huge amount of concern around these issues. While young people want to learn about technology, they are concerned about how potential future employers will use AI to control people rather than to empower them.

Sean Hinton, CEO & Founder of SkyHive Technologies, reminded us of 2016, when thought leaders around the world began talking about the Fourth Industrial Revolution, highlighting that, as in previous industrial revolutions, people have always been fearful of new technology and innovation. But the good news is that never in the world’s history has any innovation ever shrunk the labour market. As a result, the global labour economy has continued to grow, evolve and adapt to new technologies.

Susan Scott-Parker, Chief Executive of Business Disability International, pointed out, that we also need to discuss the excluded groups and those who are forced to move on to their next job after being made redundant by AI. And whilst we don’t have data accurate enough to ensure fair and equal treatment, despite knowing that 10 to 12% of the workforce of any large company is going to have a disability, we know that using AI recruitment tools that don’t allow flexibility is going to further ostracise those already marginalised. To rectify this, the developers need to be talking, consulting, and learning from people with disabilities from a wide range of backgrounds.

While Mark Keese fully agreed that AI recruitment tools could be a major barrier for people with disabilities, he also mentioned that AI could make the workplace more accessible. He called for businesses and governments to work together to provide the right incentives to develop AI applications to help people with disabilities. The OECD will publish a new study on disability and access to work in the coming weeks building on interviews with 70 or more actors in the field of disability, including AI developers and users, policymakers, and accessibility specialists.

The world of work will be completely transformed. And the transformation is going to be dramatic because AI is not only good at organising information, and analyses of various kinds. AI today is also able to perform creative tasks, increasingly better than human beings. AI is even able to form empathic relationships with people.

What the future holds

We mustn’t have a linear view of the progression of AI. Soumitra Dutta predicted that we will see exponential changes in AI systems. The world of work will be completely transformed. And the transformation is going to be dramatic because AI is not only good at organising information, and analyses of various kinds. AI today is also able to perform creative tasks, increasingly better than human beings. AI is even able to form empathic relationships with people.

While AI advances at lightning speed, according to Mark Keese, regulation tends to change at a much more glacial pace. We need to speed up this pace, but it needs to be done in an agile way – we must be careful not to entrench existing inequalities and the existing market power of some of the leading developers. We cannot wait for a generational change either – we need to look at adult learning systems and decide how to change, upskill, and reskill workers who are already in the workplace so that they have the right skills to complement and work with AI.