AI adoption in the UK labour market could lead to 7.9 million job losses and no GDP gains, according to a worst case scenario based on research from the Institute for Public Policy Research (IPPR).

The IPPR is a “progressive” think tank and registered charity funded by charitable trusts, foundations, businesses, unions and voluntary sector organisations as well as local and national government departments and individuals.

The study looked at the impact of generative artificial intelligence (AI) on the UK. It identified two key stages of generative AI adoption: the first wave, which is here and now, and a second wave in which companies will integrate existing AI technologies further and more deeply into their processes.

The Institute’s analysis of 22,000 tasks in the UK economy, covering every type of job, found that 11% of tasks done by workers are already exposed to in the first wave. It identifies ‘routine cognitive’ tasks (such as database management) and ‘organisational and strategic’ tasks (such as scheduling or inventory management) as most exposed to generative AI, which can both read and create text, software code and data.

Meanwhile, the study also showed that this could increase to AI doing 59% of tasks in the second wave. This would also impact non-routine cognitive tasks (such as creating and maintaining databases) and would affect increasingly higher earning jobs.

It says that back-office, entry level and part time jobs are at the highest risk of being disrupted during the first wave. These include secretarial, customer service and administrative roles.

The Institute noted that women are more likely to be in such jobs, which means they will be among the most affected. Young people are also at high risk as firms hire fewer people for entry-level jobs and introduce AI technologies instead. In addition, those on medium and low wages are most exposed to being replaced by AI.

There are three illustrative scenarios for the potential impact of the second wave of AI adoption on the labour market, depending on policy choices:

• Worst case scenario: full displacement: all jobs at risk are replaced by AI, with 7.9 million job losses and no GDP gains
• Central scenario: 4.4 million jobs disappear, but with economic gains of 6.3% of GDP (£144 billion per year)
• Best case scenario: full augmentation: all jobs at risk are augmented to adapt to AI, instead of replaced, leading to no job losses and an economic boost of 13% to GDP (£306 billion per year)

The Institute also modelled three scenarios for the potential impact of “here and now” generative AI on the labour market:

• Worst case scenario: full displacement: 1.5 million jobs are lost, with no GDP gains
• Central scenario: 545,000 jobs are lost, with GDP gains of 3.1% (£64 billion per year)
• Best case scenario: full augmentation: no jobs are lost, with GDP gains of 4% (£92 billion per year)

The report also noted that wage gains for workers could be up to more than 30% in some cases, but they could also be nil.

Deployment of AI could also free up labour to fill gaps related to unaddressed social needs. For instance, workers could be re-allocated to social care and mental health services which are currently under-resourced. The modelling shows that there is no single predetermined path for how AI implementation will play out in the labour market. It also urges intervention to ensure that the economic gains are widely spread, rather than accruing to only a few.

Carsten Jung, senior economist at IPPR, said, “Already existing generative AI could lead to big labour market disruption or it could hugely boost economic growth, either way it is set to be a game changer for millions of us. Many firms are already investing in it, and it has potential to speed up many more tasks as more businesses adopt it.”

Jung added, “Over the next five years it could transform knowledge work. The question now is less whether AI can be useful, but rather how fast and in what manner employers will use it. History show that technological transition can be a boon if well managed, or can end in disruption if left to unfold without controls. Indeed, some occupations could be hard hit by generative AI, starting with back office jobs.”

“But technology isn’t destiny and a jobs apocalypse is not inevitable – government, employers and unions have the opportunity to make crucial design decisions now that ensure we manage this new technology well. If they don’t act soon, it may be too late,” Jung said.

The Institute added that without government action and with companies left to their own devices, the worst-case scenario is a real possibility. It recommends the government develops a job-centric industrial strategy for AI that encourages job transitions and ensures that the fruits of automation are shared widely across the economy.

According to the study, this should include supporting green jobs, as green jobs are less exposed to automation than non-green jobs; fiscal policy measures, such as tax incentives or subsidies to encourage job-augmentation over full displacement; and regulatory change, to ensure human responsibility of key issues, such as with health.

Bhargav Srinivasa Desikan, senior research fellow at IPPR, said, “We are at a sliding doors moment, and policy makers urgently to develop a strategy to make sure our labour market adapts to the 21st century, without leaving millions behind. It is crucial that all workers benefit from these technological advancements, and not just the big tech corporations.”