Why The Washington Post is building chat bots to deliver news

When Microsoft’s artificially intelligent bot Tay went rogue and was taught to be racist, publishers got a free lesson in what can go wrong with bots, Garett Sloane writes. Despite the risks, The Washington Post is building a news bot. Director of product Joey Marburger says the Post is managing those risks in how the bot is being developed: “We’re not going to have an algorithmic conversation learning bot that starts to create its own identity from nowhere, so we avoid that Microsoft Tay debacle. But we do want it to have a personality and tone, so we will give it that. That should really be the tone and personality of the Washington Post, to a certain extent. So it’s got to be able to handle some basic conversation, but not in the way of tricking it into denying the Holocaust.”