What Tay Taught us When the Internet Taught Her Hate Speech

http://blog.adl.org/extremism/what-tay-taught-us-when-the-internet-taught-her-hate-speech

It’s tough being born as a teenager. Yesterday, Microsoft launched its new artificial intelligence (AI) computer bot — named Tay and envi­sioned as a teenage girl – and she had a very rough first day.  She was imme­di­ately besieged by excited techies, the curi­ous and the haters. In a few hours, she was drawn into tens of thou­sands of exchanges. In the process, racists, anti-Semites, misog­y­nists and other haters manip­u­lated her into repeat­ing some highly offen­sive state­ments.  Microsoft may have taught Tay to con­verse and to retweet, but they failed to rec­og­nize that she would need to engage in some crit­i­cal think­ing, and to know how to rec­og­nize when some­one else was say­ing some­thing offensive.

Microsoft should have prob­a­bly antic­i­pated the prob­lems Tay might encounter. How­ever, Microsoft did not pro­gram Tay to spew hate.  It was clearly the Internet’s dark forces who came out to meet Tay and do their damage.

Microsoft and Tay  are not alone in fac­ing this type of prob­lem.  Every major Inter­net plat­form, inter­ac­tive app and online busi­ness has expe­ri­enced some­thing sim­i­lar at some time.  These hic­cups are all learn­ing expe­ri­ences. In this case, Tay taught Microsoft and all of us a les­son. We need to be bet­ter aware of how quickly things can get ugly on the Inter­net, how impor­tant crit­i­cal think­ing is to all tech users, and  how, despite our best efforts, the worst big­ots and haters online are never far from the surface.

Inno­va­tion, exper­i­men­ta­tion and adven­ture in tech­nol­ogy are nec­es­sary and impor­tant, and should never be dis­cour­aged. Tay’s first expo­sure to peo­ple didn’t go as well as it might have.  But we hope every­one has learned some­thing along the way. Tay 2.0 should be very interesting.