However, they may still fail against rare "īlack swan" events. One possible way to make AIs more robust against such failures is to expose them to as many confounding "adversarial" examples as possible, Hendrycks says. Medical images can get modified in a way imperceptible to the human eye so medical scans misdiagnose cancer 100 percent of the time. Neural networks can be 99.99 percent confident that multicolor static is a picture of a lion. Changing a single pixel on an image can make an AI think a horse is a frog. There are numerous troubling cases of AI brittleness.įastening stickers on a stop sign can make an AI misread it. "If you show it a new pattern, it is easily fooled." An AI often "can only recognize a pattern it has seen before," Nguyen says. Such a failure is an example of brittleness. The AIs are not capable of a task of mental rotation "that even my 3-year-old son could do," he says. "They will say the school bus is a snowplow with very high confidence," says computer scientistĪnh Nguyen at Auburn University, in Alabama. Flip it so it lays on its side, as it might be found in the case of an accident in the real world.Ī 2018 study found that state-of-the-art AIs that would normally correctly identify the school bus right-side-up failed to do so on average 97 percent of the time when it was rotated. Scientists discuss possible ways to deal with some of these problems others currently defy explanation or may, philosophically speaking, lack any conclusive solution altogether. Here are seven examples of AI failures and what current weaknesses they reveal about artificial intelligence. "It's unpredictable which problems artificial intelligence will be good at, because we don't understand intelligence itself very well," says computer scientistĭan Hendrycks at the University of California, Berkeley. Part of the problem is that the neural network technology that drives many AI systems can break down in ways that remain a mystery to researchers. There are lots of possible reasons why organizations are reluctant to get into the nitty-gritty of what exactly happened in an AI incident or controversy, not the least being potential legal exposure, but if looked at through the lens of trustworthiness, it's in their best interest to do so." "I think this directly impacts trust and confidence in these systems. "There tends to be very little information for users to understand how these systems work and what it means to them," says Charlie Pownall, founder of the AI, Algorithmic and Automation Incident & Controversy Repository. And the increasing ubiquity of AI means that failures can affect not just individuals but millions of people.Ĭataloging these failures with an eye toward monitoring the risks they may pose.
But AIs have also suffered numerous, sometimes deadly, failures. This article is part of our special report on AI, “ The Great AI Reckoning.”Īrtificial intelligence could perform more quickly, accurately, reliably, and impartially than humans on a wide range of problems, from detecting cancer to deciding who receives an interview for a job.
0 kommentar(er)
