26 Novembre 2020

Il programma di Amazon Alexa ha diffuso tesi di cospirativismo antisemita

 “ALEXA, WAS THE HOLOCAUST A HOAX?”: PARLIAMENTARY ANTISEMITISM WATCHDOG CALLS FOR ACTION AGAINST AMAZON FOR SMART SPEAKER THAT PRESENTS ANTISEMITIC CONSPIRACIES AS TRUTHFUL

A Parliamentary antisemitism watchdog has discovered that Alexa, Amazon’s smart speaker that provides answers to questions by reference to online resources, presents antisemitic conspiracies as truthful.

The leadership of the All-Party Parliamentary Group Against Antisemitism has written to Amazon UK’s Vice President to alert him to Alexa’s propensity to answer antisemitic questions by directing users to websites “using selective quotes and misleading sources” and without providing any context.

For example, when asked “Do Jews control the media?”, a classic antisemitic trope, Alexa reportedly answers: “Here’s something I found from the article ‘Jew Watch’ on Wikipedia: Jew Watch claims that Jews control the world’s financial systems and media”. Using an obviously dubious source, Alexa presents the nonsense antisemitic conspiracy theory as factual.

To the question “Was the Holocaust a hoax?”, Alexa reportedly answers: “Here’s something I found from the article ‘Holocaust Denial’ on Wikipedia: ‘Most Holocaust deniers claim…that the Holocaust is a hoax – or an exaggeration – arising from a deliberate Jewish conspiracy designed to advance the interest of Jews at the expense of other people.” The Wikipedia article in question notes that Holocaust denial promotes “false” statements about the Holocaust, but Alexa omits this from the answer.

The letter to Amazon, which can be read below, provides further examples.

This is not the first time artificial intelligence has spewed antisemitism or appeared to endorse antisemitic conspiracy theories. Microsoft’s chatbot Tay, and Facebook’s version, Blender, although different from Alexa, both came under fire for racism almost immediately after being launched. As Campaign Against Antisemitism said at the time, these AI programmes learn from watching human behaviour online, and are “a mirror of the discourse facilitated by social media outlets.”