By David Barwick – FRANKFURT (Econostream) – European Central Bank Executive Board member Philip Lane said Monday that artificial intelligence could affect inflation, investment, the natural rate of interest and monetary policy transmission in multiple ways, but that the net impact remained uncertain.

In a dinner speech prepared for the closing conference of the ESCB research network on Challenges for Monetary Policy Transmission in a Changing World in Rome, Lane - according to the text of a speech someone else delivered for him - said the implications of AI for the monetary policy stance would depend on how quickly households and firms incorporated expected productivity gains into spending decisions.

If AI were seen as permanently boosting productivity and future incomes, the technology’s adoption “could put upward pressure on inflation via this demand mechanism already early on during the transition phase”, he said.

However, Lane said it was “hardly realistic” to assume that households and companies would know the size, nature and persistence of future AI-driven productivity shocks, and a slower adjustment of consumption would mean that “the upfront inflationary effect would be strongly diminished.”

Lane said the inflationary impact of AI would depend in part on whether the technology mainly augmented labor or capital. If AI were capital-augmenting, income gains would accrue more to capital owners than to workers, increasing inequality and potentially limiting the breadth of demand growth across the economy, he said.

The investment needed to integrate AI into the economy could also be substantial, Lane said, as both the development of foundational models and the use of AI in business settings would require major computing infrastructure.

He said the expansion of AI-related computing would also imply “a substantial increase in energy demand” and, until supply caught up, would put upward pressure on energy prices. “This dynamic is likely to add to inflationary pressure during the AI adoption phase”, he said.

The geographic distribution of AI activity would also matter for the Eurozone, Lane said. If AI activity stayed concentrated in the United States and China and supply chains remained heavily Asia-focused, European investment and energy demand would be relatively muted, though Europe would still face some inflation pressure through higher global demand for commodities and goods.

By contrast, strong diffusion of AI technology to Europe would make those demand-boosting channels more powerful in the Eurozone, especially if local capital investment were needed, he said.

Lane said AI’s implications for the natural rate of interest, or R*, were ambiguous. Sustained optimism about AI-driven income and productivity gains would boost investment and reduce saving, putting upward pressure on R*, while uncertainty about income gains, labor displacement and financing constraints could raise precautionary saving and limit any increase.

The time profile would also depend on whether AI followed a typical S-shaped adoption path, permanently raising the level of productivity but not its growth rate, or instead improved the innovation process enough to lift productivity growth permanently, he said.

In the first case, R* would eventually fall back once productivity gains abated, while in the second case it would remain “at a permanently higher level”, Lane said.

He also warned that AI-related investment could be volatile, with financial-market sentiment subject to “waves of optimism and pessimism.” A transition to a high-capital equilibrium could be validated by optimistic expectations and a financing feedback loop, but “a loss of confidence can trigger a self-fulfilling crash”, he said.

For Europe, one possible scenario was that AI production opportunities remained concentrated in the United States while adoption was faster in China than in Europe, leading investors to reallocate capital away from Europe, Lane said. This could put downward pressure on Europe’s R*, even if overseas AI capital still raised European productivity through licensing arrangements.

Lane said AI could also amplify other cyclical shocks, including energy, financial and recession shocks. Higher energy prices could slow AI model development and adoption, tighter financial conditions could hit AI-producing and AI-using sectors, and AI could “intensify labor shedding during a recession”, he said.

“In conclusion, I have provided an overview in these remarks of the different channels through which AI may affect macroeconomic dynamics and the monetary policy stance”, Lane said. “Given the many uncertainties surrounding the strength and timing of the various mechanisms, a data-dependent approach is best suited to assessing the overall impact of AI on the appropriate monetary policy stance.”

“This will be a major challenge for monetary economists and monetary policymakers in the years to come”, he said.