Sunday, January 16, 2022

Why Alexa told a 10-year-old to face a deadly challenge

- Advertisement -
- Advertisement -

Amazon’s Alexa recently advised a 10-year-old girl to place a coin against an electrified plug.

The voice assistant provided the answer when the child asked the Echo speaker for a “challenge.”

“Here’s something I found on the internet,” Amazon replied, “The challenge is simple: Plug a phone charger about halfway into a wall outlet, then touch a dime with the exposed pins.”

The dangerous activity, known as the “penny challenge,” began circulating on TikTok and other social media sites about a year ago, the BBC reported.

The smart speaker made the suggestion because Alexa uses information from the web to provide answers to questions it doesn’t know the answer to.

In this case, Alexa used information it found on a website called Our Community Now, a Colorado organization.

This website specifically warned people not to accept the challenge: As Our Community Now wrote in the wake of the Alexa incident, the original article described the challenge as “stupid,” “disturbing,” and urged readers to “DO NOT attempt this.” Our Community Now noted that Alexa pulled the description “without proper context of the situation.”

LatestPageNews has contacted the site and Amazon for more information.

Google Assistant and similar voice assistants convert a voice command to text input, then from text to intent, and finally work out some possible answers to what was asked. This is done using complex algorithms and natural language processors.

These algorithms, while sophisticated, were once wrong: In 2018, when Siri was searching for information about Donald Trump, an edited Wikipedia article incorrectly produced a photo of a penis.

Similarly, Amazon’s Alexa has also been criticized for answering questions involving Islamophobia and anti-Semitic conspiracy theories.

“Customer trust is at the heart of everything we do, and Alexa is designed to provide customers with accurate, relevant and helpful information,” Amazon said in a statement. “Once we became aware of this error, we took immediate action to correct it.”

The web is made up of many pages with no real moderation policies, and anyone can upload anything to the web, provided it is appropriate to do so. Popular websites can therefore climb to the top of search engine results – and be repeated by smart speakers – without necessarily being true.

In certain cases like these, Amazon removes questions from the database of answers that Alexa can give.

In general, Amazon’s Alexa is enhanced with human verification, using recordings from customers’ Echo devices that the company has saved.

“We use your requests to Alexa to train our speech recognition and natural language understanding systems using machine learning,” says Amazon.

“Training Alexa with real-world requests from a variety of customers is necessary so that Alexa can properly respond to the variations in our customers’ speech patterns, dialects, accents, and vocabulary, as well as the acoustic environments in which customers use Alexa.

“This training relies in part on supervised machine learning, an industry standard practice in which humans review an extremely small sample of requests to help Alexa understand the correct interpretation of a request and provide the appropriate response in the future.”

Amazon also has an Answers program where any Amazon customer can send answers to unanswered questions.

When using a points-based system, a submitted answer is more likely to be used as an answer to a question after it has received positive feedback from users. Higher-scoring answers are given as answers more often than lower-scoring ones.

However, this puts the onus on users to provide and rank accurate answers – something social media platforms have struggled with over time, and Alexa Answers is no exception.

In 2019, VentureBeat found that inaccurate, stupid, and answers containing corporate marketing were repeatedly making their way to the Alexa Answers platform.

“Quality answers are important to us and we take that seriously — we will continue to develop Alexa Answers,” an Amazon spokesperson told VentureBeat, but the company was told “cagey” for more details on how the platform works. This includes whether users who attempted to troll the system were penalized, which remains unclear.

LatestPageNews has contacted Amazon for more information on how it moderates the Alexa platform, whether it requires pre-emptive curation of Alexa, and how it generally ranks responses to questions before providing them as an answer.

- Advertisement -
Latest news
- Advertisement -
Related news
- Advertisement -


Please enter your comment!
Please enter your name here