Deep learning tools raise worrying questions, because they solve problems in ways that humans can’t always follow. If the tools are inscrutable bits of computer code — essentially black boxes — can they be trusted? Among researchers, there’s a growing call to clarify how deep learning tools make decisions.
↧