Report: AI Trading May Hasten a Stock Market Crash

John Lister's picture

Artificial intelligence designed to make investing more reliable could backfire in the next stock market crash, according to a global business site. The article for website "Quartz" says AI systems might not react well if market conditions suddenly change.

The problem is with the increasingly sophisticated way that computer programs can make investment decisions on behalf of human investments. Computers have been used for such processes for many years - but until recently, were simply set to follow basic preprogrammed rules. For example: the program might place a buy order if a historically strong stock dropped briefly and then began rebounding.

The key was that such programs only followed rules produced by humans. Today, many programs use artificial intelligence either to make decisions or give human investors recommendation. The artificial intelligence doesn't just follow rules, but learns and refines its own guidelines about how stocks behave and when it is most likely a good time to buy or sell a particular stock.

Market Abnormally Stable

Writers Helen Edwards and Dave Edwards point out a possible major flaw in this set-up: the systems largely use market data from recent years, simply because the amount of data collected and analyzed has grown dramatically during that time. (Source: qz.com)

The problem is that the past few years have been historically stable for financial markets: while there have been ups and downs, most stocks and the markets as a whole have changed prices at a slow, steady rate rather than rocketing or plummeting. There ars several reasons for that, but a key one is that it's cheap for companies to borrow. That means that if they do get into short-term trouble, investors don't panic because they know the company can ride it out.

Panic Cycle Could Ensue

This means that the artificial intelligence might assume these are 'normal' market conditions and simply not know how to react effectively if and when the market next crashes. That even creates the risk that the systems don't react 'rationally,' and instead panic more than a human would, starting a vicious cycle in the market.

Quartz notes that artificial intelligence coming up against unknown situations is something that is unavoidable and not necessarily a disaster; the key is whether it is designed to be able to adapt and have fail-safes if things are going wrong. For example: in 2010, the stock market crashed dramatically for half an hour with many people blaming it on automated trading systems getting caught in an extreme cycle of selling off. (Source: montrealgazette.com)

What's Your Opinion?

Do you think Quartz makes a valid argument? Would you knowingly use an investment analyst that relies partly or solely on artificial intelligence? Should there be more rules over the use of automated investment to protect the market as a whole, or is it literally "the free market at work"?

Rate this article: 
Average: 5 (4 votes)

Comments

gmthomas44_4203's picture

As you say, it depends on how the AI is set up to identify conditions.
I would welcome a crash, as that is the best time to buy! When everyone is "lemming" selling.
The first tenet of What's-his-name, "I buy when everyone is selling and I sell when everyone is buying"!