LONDON - ‘AI Says Men Are Lazy,’ a Bloomberg headline proclaimed last year. This assertion by Artificial Intelligence may be a case of the pot calling the kettle black, however, since AI readily assumes the worst traits of its human handlers, whether bigotry, deceit, or cheating, and indolence is apparently no exception to this. Slothful AI can be either harmful or beneficial depending on the scenario in which it deploys, however.
The use of AI in healthcare may well enhance diagnostic efficiency, but a University of Washington study found that one AI application resorted to heuristic shortcuts rather than actual medical pathology in diagnosing COVID-19, with potentially dire consequences for patients.
Researchers from UW’s Paul G. Allen School of Computer Science & Engineering checked chest X-rays used to detect COVID-19, per their study. The team discovered that the AI gave greater weight to certain datasets than key medical factors when analyzing whether a patient had been infected with the virus. The researchers characterized these shortcuts as the AI being “lazy.”
“AI finds shortcuts because it is trained to look for any differences between the X-rays of healthy patients and those with COVID-19,” the team said, GeekWire reported on June 3, citing the study, whose results appeared in the peer-reviewed online AI journal Nature Machine Intelligence on May 31. “The training process doesn’t tell the AI that it needs to look for the same patterns that doctors use, so the AI uses whatever patterns it can to simply increase accuracy of discriminating COVID-19 from healthy,” the researchers added.
When doctors use chest X-rays for a COVID-19 diagnosis, they already have such patient information as exposure and medical history but anticipate fresh information from the X-ray.
“A phThe content herein is subject to copyright by The Yuan. All rights reserved. The content of the services is owned or licensed to The Yuan. Such content from The Yuan may be shared and reprinted but must clearly identify The Yuan as its original source. Content from a third-party copyright holder identified in the copyright notice contained in such third party’s content appearing in The Yuan must likewise be clearly labeled as such.