Amazon today opened up its Alexa Answers program, which lets humans supply answers to queries that the digital brain can’t yet comprehend, to everyone.
Alexa Answers first launched late last year by invitation only, following an internal test that helped add 100,000 answers to Alexa’s knowledge base. Now, Amazon customers will be able to pick areas they are passionate about and help Alexa build expertise.
After signing in on the Alexa Answers website, customers can see a list of questions and apply filters like, frequently asked questions and new questions. Amazon is gamifying the experience by awarding badges and points for answers that go toward monthly and weekly leaderboards.
Building up a robust question and answer database for Alexa is an important initiative as it competes with other digital assistants. One big advantage the Google Assistant has over Alexa is the backing of Google’s market-leading search engine. By filling out Alexa’s knowledge base, Amazon can create from scratch a crowdsourced, voice-activated search engine.
However, the technology world is filled with cautionary tails of people corrupting efforts to make machines smarter. YouTube, Reddit, Facebook and many more have struggled to stop disinformation and hostile behavior for years.
You don’t even have to leave Amazon’s home region of Seattle to find a high-profile example. Two years ago, Microsoft shut down its millennial-minded chatbot experiment Tay within just 24 hours after the bot expressed racist and inappropriate leanings to some users.
Amazon put several safeguards in place to make sure inaccurate or offensive content doesn’t slip into Alexa’s dialog. Amazon has automated filters to weed out offensive questions from appearing on the Alexa Answers website and prevent profanity and other offensive language from showing up in answers.
Fast Company reports that Amazon will deploy some human editors and algorithms for quality control. There is a community element as well, according to Amazon, as Alexa customers will be able to up and down vote answers as they hear them. Answers that receive a lot of down votes will be removed.
“We’re leaning into the positive energy and good faith of the contributors, and we use machine learning and algorithms to weed out the noisy few, the bad few,” Bill Barton, Amazon vice president of Alexa Information, told Fast Company. “But we’re not going to suppress the magical experience we can give to 99 customers because one person had something different in mind.”
For now, Amazon is not requiring user-supplied answers to cite where information came from. However, Alexa will note that the answers are “according to an Amazon customer,” per Fast Company.
