It's hard to write robust software: computer programs are notoriously incapable of handling new, unforeseen situations. One of the primary purposes of artificial intelligence is to develop techniques that allow software to be as flexible as the human brain, to be able to make reasonable decisions in confusing circumstances with incomplete information. There are few more complex and confusing domains than financial markets, so it's not surprising that there are glitches in financial AI programs.
NEW YORK (Reuters) - Computers, unable to see the mold on an outdated UAL Corp (UAUA.O: Quote, Profile, Research, Stock Buzz) article, sparked confusion, a mass sell-off, and an emergency trading halt earlier this week that highlights the pitfalls of increasingly automated financial markets.Shares in the parent of United Airlines plunged about 76 percent after a nearly 6-year-old news story on its 2002 bankruptcy filing appeared online on Monday.
UAL, which is no longer in bankruptcy, scrambled for a retraction and its stock later bounced back -- but for some investors, especially frequent traders, the damage was done.
The use of algorithms -- which allow computers to make decisions in fractions of a second -- appears to be a main culprit in the UAL case.
Experts said the automated programs were applied to both the reading of the outdated news story and the trading of shares based on that information.
Humans can make disastrously bad decisions too, of course, so we shouldn't be too quick to judge our artificial counterparts! The more intelligent our software becomes, the more prone it will be to human-like errors. Even if we do ever achieve human-level intelligence in a machine, there's no reason to believe it will be any more perfect or rational than our own intelligence.
(HT: Nick.)
