OK, it is addressed in the article...
He's specifically talking about the use of AI in finance, and that an algorithm that runs amok in a particular sector:
in the after action reports people will say 'Aha! There was either one data aggregator or one model . . . we've relied on.' Maybe it's in the mortgage market. Maybe it's in some sector of the equity market
I'll throw out a microeconomic example. About a year into the pandemic, the price of used cars started going up... a LOT... in a short time. One of the reasons for the sudden changes in used car prices was that major used car resellers were using algorithms to set buying and selling prices for cars. While supply chain pressure on the new car market was unprecedented, and it trickled down to used cars, a facilitating cause is that the used car price-setting algorithms didn't really have any humans in the chain checking to see if the numbers they were kicking out made a lick of sense.
So you had companies like Carmax and Carvana buying used cars for $X, and then a month later 5X, then a month later 10X, because they were programmed to just up the offering price until they reached target stock levels. Sometimes they were buying 3+ year old used cars for more than the current price of NEW cars of similar trim level. Carvana's numbers got so whacked that it nearly sunk the company.
Now imagine that kind of a runaway algorithm in stocks, bonds, real estate, etc. It's 2008 all over again.