A mass experiment aimed at improving Amazon Alexa's communications skills has had some unintended side-effects.
With approximately 43 million people using Echo speaker devices and the company's chipper digital assistant, Alexa, Amazon has embarked on a strategy to improve its communications skills. To that end, the annual $500,000 Alexa Prize was created in 2016 with the aim of encouraging computer science students to help the digital assistant become more helpful - and more human, reports Reuters.
At stake is a fast-growing market for gadgets with virtual assistants. An estimated two-thirds of U.S. smart-speaker customers, about 43 million people, use Amazon’s Echo devices, according to research firm eMarketer. It is a lead the company wants to maintain over the Google Home from Alphabet Inc and the HomePod from Apple Inc.
Over time, Amazon wants to get better at handling complex customer needs through Alexa, be they home security, shopping or companionship.
To make that happen, the company in 2016 launched the annual Alexa Prize, enlisting computer science students to improve the assistant’s conversation skills. Teams vie for the $500,000 first prize by creating talking computer systems known as chatbots that allow Alexa to attempt more sophisticated discussions with people. -Reuters
The results of Amazon's adventures in AI have been... interesting, to say the least.
Owners of Amazon Echo devices can participate in the experiment by saying "let's chat" to their devices - at which point Alexa will tell the user that one of the bots will take over - removing conversational constraints. Between August and November of this year, the three bots which made it to the Alexa Prize finals had 1.7 million conversations, according to Amazon.
The project has been important to Amazon CEO Jeff Bezos, who signed off on using the company’s customers as guinea pigs, one of the people said. Amazon has been willing to accept the risk of public blunders to stress-test the technology in real life and move Alexa faster up the learning curve, the person said.
The experiment is already bearing fruit. The university teams are helping Alexa have a wider range of conversations. Amazon customers have also given the bots better ratings this year than last, the company said. -Reuters
That said, Alexa has had its share of gaffes - leading Amazon to shut down at least one bot according to three people familiar with the matter.
One user was shocked when Alexa blurted out "Kill your foster parents," calling the incident "a whole new level of creepy." Amazon investigated the incident and found that the bot had quoted a Reddit post without context.
Another Alexa bot gave a lecture on sex acts, as well as a discourse on dog defecation.
During last year’s contest, a team from Scotland’s Heriot-Watt University found that its Alexa bot developed a nasty personality when they trained her to chat using comments from Reddit, whose members are known for their trolling and abuse.
The team put guardrails in place so the bot would steer clear of risky subjects. But that did not stop Alexa from reciting the Wikipedia entry for masturbation to a customer, Heriot-Watt’s team leader said.
One bot described sexual intercourse using words such as “deeper,” which on its own is not offensive, but was vulgar in this particular context.
“I don’t know how you can catch that through machine-learning models. That’s almost impossible,” said a person familiar with the incident. -Reuters
In order to combat the potty-mouthed bots, Amazon has rolled out tools to filter profanity and sensitive topics, while constantly monitoring chat logs to be on watch for conversations that are too spicy. That said, the company admits that it cannot anticipate every potential problem, which means Alexa could find entirely new ways to shock users.
"We are mostly reacting at this stage, but it’s still progress over what it was last year," according to Amazon's VP and head scientist of Alexa AI, Rohit Prasad.