Substitute 'education' for 'health care' and the same argument applies. We want explanations for decisions or diagnoses made by artificial intelligence; this requirement shows up in any number of documents on the ethical use of AI. But we should be cautious in making this demand, according to this article. "This consensus, at least as applied to health care, both overstates the benefits and undercounts the drawbacks of requiring black-box algorithms to be explainable." The article is behind a stupid paywall, but it's short and can be viewed on Twitter in full here. See also Eric Topol's full thread. Image: Explanation ontology (full size version).
Today: 5 Total: 92 [Direct link] [Share]
