"same [AI] techniques will be able to … transform whole industries. But this won’t happen—or shouldn’t happen—unless we find ways of making techniques like deep learning more understandable to their creators and accountable to their users."https://www.technologyreview.com/s/604087/the-dark-secret-at-the-heart-of-ai/?utm_source=facebook&utm_medium=tr_social&utm_content=2017-12-31&utm_campaign=site_visitor.unpaid.engagement&fbclid=IwAR26Ii8D7xgNZdxyg4pS2qI5-Ad-I9WQdVYIQ06nIpSVPeAHN31uVg1IQpU …
-
Show this thread
-
Er, consider credit scores, which few customers or regulators understand. Our whole economy is full of stuff most people don't understand. Why hold future AI to a higher standard than we've held everything in the past?
6 replies 4 retweets 39 likesShow this thread -
Replying to @robinhanson
This is a practical problem, in that enterprise customers insist on visibility into how the algorithm works. I built such a tool for our client at Palantir; it was just a random forest, but we had to show the humans what factors contributed to the algorithm's decision.
2 replies 0 retweets 0 likes
Many industries are TOTALLY unused to rubrics that are probabilistic and multifactorial. Not an 'AI' issue but a statistics issue. Many people are uncomfortable with uncertainty. They won't accept a rubric that spits out a probability without a lot of skeumorphism.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.