A study by the Economist Intelligence Unit (EIU) for Google has predicted that the UK will fall behind other leading economies unless the government makes radical changes to support AI (artificial intelligence) and data sharing.
The EIU Risk and Rewards report warned that the UK is expected to underperform compared with other countries in the analysis, in terms of both GDP and productivity growth.
Compared with the US, Australia, Japan, Asia and Developing Asia, the UK has the most to lose from poor public policy, said the report, with falls of 1.83% in GDP and 1.79% in productivity.
“For a number of potential reasons, including Brexit and the diminution of the highly productive financial services sector, the baseline forecast is for the country’s productivity to be negative between now and 2030,” the report said.
Between 2016 and 2030, productivity in the UK (measured in billions of dollars per million man-hours worked) is expected to decline from 0.045 to 0.041, with a compound annual growth rate (CAGR) of -0.62%.
The EIU’s rationale for the UK’s poor performance is based on falling productivity in line with recent performance. The country’s struggle with poor productivity has long been a top concern for economic policymakers, the EIU said.
The report presented three possible scenarios for AI adoption. Even in the best case, where AI complements human productivity – Scenario 1 in the EIU report – the UK’s growth in terms of GDP remains stagnant while the US, by comparison, grows at a CAGR of 1.59%, the report said.
In Scenario 2, which presents a more strategic and policy-driven model for incentivising AI research and adoption, the UK’s GDP is set to grow the least of all countries assessed in the report.
Read more about artificial intelligence
- Artificial Intelligence is red hot. But what lies beyond the hype? Once it was big data, then cloud, now it is artificial intelligence, and that subset of it which is machine learning.
- Artificial intelligence and blockchain initiatives are earmarked for a doubling of investment, finds the Computer Weekly/TechTarget IT Priorities survey for 2018.
Overall, the EIU forecast that if the UK does not take enough action to support AI initiatives, the country’s economy will be US$420bn smaller in 2030 than it was in 2016.
At the World Economic Forum last month, prime minister Theresa May reiterated plans announced in the government’s industrial strategy to create a national research centre for AI to foster and retain research talent. May said the key to harnessing technology the right way is to ensure people are equipped with the skills they need to use digital technologies, and ensure industry and government work together on this. “We can’t sit back and leave it to the labour market,” she said.
The EIU warned that labour costs will affect AI investment. Quoting projections from the 2016 OECD, the report said 9% of jobs will be completely taken over by technology in the next two decades, with a further 25% having around half their tasks automated.
In the report, Jerry Kaplan, futurist and fellow of the Stanford Center for Legal Informatics, said: “Whether you make someone more productive or replace them completely, you can do more work with fewer workers. Both put people out of work, but grow the economy.”
According to Kaplan, AI can lead to a new age of affluence and leisure, but reaching this state will depend on addressing labour volatility and income inequality. “There will be plenty of jobs around, but the main effect of technology will be to change the skills that people need to accomplish their jobs,” he said.
Jin-Hyung Kim, emeritus professor at the Korea Advanced Institute of Science and Technology, said: “The best benefit of AI is automation, but full automation is not a good story for humans.”
Kim said promoting and providing high-quality public data is the single most important policy for AI at present and in the near future.
Given that AI runs on the ability to “learn” from new data, the EIU warned: “Should governments fail to develop and enact policies that assuage privacy and protection concerns while ensuring access to data, the growth potential of AI will be diminished.”
Reasonable debate needed
Chris Clague, the editor of the EIU report, said: “The debate over the impact of machine learning and artificial intelligence is an important one and, like all important debates, it needs to be reasonable and informed.”
According to the EIU, policymakers face a number of important choices with regard to machine learning and its impact. Chief among these is investing in skills and education, and not just science, technology, engineering and maths (Stem) skills. The demand for “soft skills,” such as team building and critical thinking, is set to rise, which means technical education and training alone will not help economies to cope with the churn in labour markets that machine learning is likely to cause.
Getting policy right on data and investing in research and development (R&D) and technology will also be critical. Citizens’ concerns about privacy and the security of personal data need to be assuaged so that data can continue to flow within and between countries. The public sector also needs to return to investing in R&D so that it isn’t just the private sector that is advancing technology.
Without the fillip of new tax incentives for private sector investment, machine learning technology continues to develop in line with base case assumptions, which, for the UK, forecasts GDP to grow by 0.63% by 2030, compared with a baseline of 1.84% GDP growth in the US.
The EIU warned that if UK policymakers fail to step up to the economic challenges posed by machine learning technology, the dominant impact of such technology will be to replace workers, rather than increase their productivity.
The lack of government support for national data-sharing reduces the overall positive economic impact of machine learning, which requires data of high quality and quantity to be of maximum usefulness. Most crucially, a lack of investment in upskilling the workforce will result in the substitution of human work by automation being stronger than the complementary effect, whereby AI augments human work.