(Bloomberg Opinion) — Central bankers aren’t born as chronic worriers, but they quickly acquire the trait. They are now spending considerable time fretting about artificial intelligence: Its ability to play havoc with prices, jobs, and the security of banking. As gut-wrenching as the meltdown of 2008 was, imagine if a rogue machine turbocharged a market rout. 

It’s not quite a doomsday scenario where AI runs amok and destroys the Earth, said Eddie Yue of the Hong Kong Monetary Authority at a recent conference. But there are plenty of dangers emerging, he added. Yue’s counterpart in Singapore warned of the potential for fraud and cyber attacks. American and UK officials are fearful that algorithms will be used to curtail lending to minorities. While acknowledging the benefits of rapid technological advances to the overall economy, most are wary. 

One thing the lords of finance shouldn’t stress about is dilution of their power. Sure, the legions of Ph.D economists that staff central banks may thin. New algorithms that sift real-time data on everything from car sales to foot traffic at malls will rightly push analysts to think about how their roles will transform. But rather than make the men and women who actually set interest rates redundant, AI could make them mightier citizens. 

The Bank for International Settlements declared as much, saying that the most basic of tasks, deciding borrowing costs, will still be done by mortals. HAL, the computer that assumes God-like qualities in the film 2001: A Space Odyssey, isn’t coming for the Federal Open Market Committee and its global peers. “The ways we organize ourselves and our societies are that we like to hold human beings accountable,” Cecilia Skingsley, head of the Innovation Hub at the BIS, told reporters last month. “You know, changing politicians, possibly changing central-bank governors from time to time.”

She may be lowballing it. The importance of Federal Reserve Chair Jerome Powell and his cohort may only grow. As retailers develop applications to keep ever closer tabs on competitors and broader markets, the price of milk in Denmark, for example, may fluctuate during a press conference by Powell, argues Lars Christensen, an associate professor at the Copenhagen Business School. When OPEC raises or cuts oil production, that’s very quickly reflected in the price of gasoline at the roadside. Why shouldn’t the same apply to basic food staples, asks Christensen, cofounder of PAICE, a consulting firm specializing in AI and data analysis.

“In many high-income countries, we already have electronic price tags,” he told me. “You might as well plug them into an algorithm. I don’t think my example of standing in a supermarket watching the price of milk change on the screen as Powell announces rates is unrealistic. For practical reasons, we might change the price only when the supermarket closes or you might have a mechanism that says the price can’t be increased while the customer is in the store. The concept is there.”(1)

Utterances from a generation ago can be resurrected to provide bond-market signals, thanks to a ChatGPT-based language model. JPMorgan Chase & Co. built a program that uses speeches dating back decades to detect the evolution of policy signals. The bank’s economists discovered that when the model showed a rise in inflation concern among Fed speakers between meetings, the following FOMC statement had grown more hawkish. The opposite is also true. Turn that into a trading strategy and the opportunities for a payday are plenty. Initially tracking the Fed, European Central Bank and Bank of England, JPMorgan expanded the method to 10 major developed-market central banks.

There’s always room for nuance and considered opinion. Sometimes the signals aren’t especially clear. For example, how do you interpret the Reserve Bank of Australia’s phrase du jour, “We aren’t ruling anything in or out?” A career Bundesbank policymaker might be less inclined to ease than, say, someone from the Bank of France. Bank of Japan Governor Kazuo Ueda can veer off on tangents. His predecessor delighted in surprising investors.   

AI works best when complementing human judgment. In some arenas, there is no substitute for experience. The machines helping reduce tax evasion in Turkey perform a public service, for example. Still, nobody would consider the country a gold standard for performance: Inflation is a stratospheric 72%. There needs to be a combination of electrons and brain waves.The employment mandate of central bankers themselves is unlikely to disappear. If Donald Trump wins this year’s presidential election, he’s pledged not to re-appoint Powell, who may not even want a third term. Safe to say HAL won’t make the shortlist —  this time.  

More From Bloomberg Opinion: 

(1) Christensen discussed the subject at length in George Mason University’s Macro Musings podcast last month.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Daniel Moss is a Bloomberg Opinion columnist covering Asian economies. Previously, he was executive editor for economics at Bloomberg News.

More stories like this are available on bloomberg.com/opinion