The Bank of England is using AI to measure the economy amid a crisis in the reliability of official data.
Huw Pill, the chief economist at the Bank of England, said the central bank had been using AI models and other new techniques, such as web scraping, to extract more details about the economy recently.
Official data sources from the Office for National Statistics (ONS) are increasingly coming into question . Response rates for key surveys plummeted during the pandemic and have remained low for years, though have improved recently.
The ONS has been forced repeatedly to revise key data.
It admitted earlier this year to a fresh error affecting labour market figures released in July.
Speaking to the Interest Journal, Mr Pill said that it was the experience during the pandemic that prompted the Bank to look to other sources of data.
He said: "Recently, we have been using AI models to draw more quantitative signals from the qualitative data coming from corporate reports, survey responses or agents' conversations with their business contacts across the country."
The Bank operates a network of agents which have one-to-one confidential conversations with businesses and community organisations across the country. They have roughly 6,000 discussions with contacts a year, which are used to help inform policymakers at the Bank about what is happening on the ground.
Mr Pill added: "Using web-scraping and other more modern technologies to build new sources of information for policy decisions lies at the heart of our strategy and we are making good progress in pursuit of this ambition.
"[The] experience during the Covid pandemic – when the face-to-face interviews underpinning traditional statistical surveys proved impossible to conduct – gave impetus to this work."
Mr Pill also responded to suggestions that the members of the Bank's monetary policy committee (MPC), who collectively determine whether interest rates rise or fall, don't think independently.
"On occasion, the MPC has been accused of 'group think' or relying on potentially stale conventional wisdom in coming to its decisions," he said. "That has not been my experience on the MPC.
"I have now voted at 40 monetary policy meetings during my tenure on the committee, and we have voted unanimously on only one occasion."
The Bank was contacted for comment.
