AI and the global economy: promises, challenges and policy solutions
Governments and global financial institutions are increasingly taking the promises and risks of AI seriously, writes Julian Jacobs, senior economist at OMFIF.
Artificial intelligence has captured the public imagination in 2023. After decades in which innovations were felt chiefly among a handful of sectors and industries, the emergence of large language models and the steady growth of chat bots have brought greater attention to the technology.
Yet AI’s transformative capacity extends far beyond LLMs. AI, and the machine learning methodologies that underlie it, comprises a large range of tools and methods, with applications across every global industry and nearly all aspects of human existence – from healthcare to education to finance.
The rise of LLMs this year has been an important development for economists and the financial sector. For decades, there had been proclamations about the potential for AI to reshape the labour market and economy. And yet, when economists and social scientists actually looked at the data on the effects of AI, they were often left asking: ‘What AI?’
Until 2023, research on the macroeconomic effects of AI was effectively speculation, or at the very least extrapolation from previous periods of significant technological change. This not only made it harder to do empirical analysis, but it also quarantined AI economic research as a space existing far away from mainstream economists. In the 2000s, scholars like David Autor and Daron Acemoglu began to change this, but economists continued to be confounded by one factor in particular: AI labour and productivity effects – the key variable they hoped to study – was largely absent from economic data and indicators, at least in a direct form.
Today, this is no longer true. Rather than the stuff of science fiction, AI is beginning to spread across industries. Early evidence – from scholars such as Erik Brynjolfsson – suggests that it is starting to bolster productivity growth in some sectors. As the rate of progress accelerates, we should expect AI to increasingly substitute labour market tasks in a growing diversity of occupations.
On some level, these developments are welcomed. Productivity growth has been sluggish in much of the West since the 1980s. AI may buck the trend of previous digital technologies having transformative effects, while not seeming to radically reshape economists’ traditional productivity benchmarks.
Yet with such promise comes risk: labour automation would displace workers in key legacy industries, like manufacturing, transportation and healthcare. This poses the risks of growing inequality and elevated market concentration. Meanwhile, widespread AI deployment in financial markets could reduce volatility, but result in convergence of investor behaviour that leads to more violent asset price swings, as Gary Gensler has warned.
It is in this context that OMFIF has chosen to focus this edition of the Bulletin on AI with a particular emphasis on finance, macroeconomic considerations and public policy. We are honoured to feature contributions from leaders and experts in AI economics, policy and safety.
Nneka Chike-Obi from Sustainable Fitch highlights the transformative potential of AI in sustainable investing, emphasising its usefulness in developing and analysing environmental, social and governance metrics while acknowledging the high volumes of energy required for AI model training. Arboath Group’s Christopher Smart brings attention to the unique nature of AI's ‘productivity boost’ in the white-collar workforce through faster and cost-effective software upgrades, leading to potential deflation.
Jai Vipra at AI Now Institute underscores the re-emergence of industrial policy in advanced economies motivated by national security. She urges a shift in focus from security to issues like unemployment and food insecurity resulting from AI in agriculture. Bilva Chandra from RAND Corporation warns of the misinformation risks of generative AI. She anticipates that AI could supercharge market manipulation, allowing malevolent entities to deceive at a scale and speed previously unseen, leading to potential macroeconomic consequences.
To counter these issues, Akash Wasil, an AI researcher, proposes three measures for regulating AI: a global cap on AI development, a potential licensing agency in the US and an international organisation akin to the International Atomic Energy Agency to oversee AI development.
Finally, experts from academic institutions share their views on the impact of AI in finance and labour markets. Eduardo Reviglio, Yale University, urges greater co-operation between the public and private sector to foster AI research and encourages industry leaders to embrace open-source collaboration to fuel intellectual property development.
Francesco Venturini from the University of Perugia cites studies suggesting that, while AI boosts company productivity and reduces labour demand, it does not significantly influence wages. Similarly, the University of Groningen’s Bart Los provides a hopeful view on the labour market implications of AI, suggesting it might save jobs much like earlier technologies did. However, he notes that while AI might revolutionise fields like software coding and website design, it could have a smaller effect on routine work.