Alexa tells 10-year-old girl to touch live plug with penny

Community
1

Amazon has updated its Alexa voice assistant after it “challenged” a 10-year-old girl to touch a coin to the prongs of a half-inserted plug.

The suggestion came after the girl asked Alexa for a “challenge to do”.

“Plug in a phone charger about halfway into a wall outlet, then touch a penny to the exposed prongs,” the smart speaker said.

Amazon said it fixed the error as soon as the company became aware of it.

The girl’s mother, Kristin Livdahl, described the incident on Twitter.

She said: “We were doing some physical challenges, like laying down and rolling over holding a shoe on your foot, from a [physical education] teacher on YouTube earlier. Bad weather outside. She just wanted another one.”

That’s when the Echo speaker suggested partaking in the challenge that it had “found on the web”. [SOURCE]

Not strictly GPT-related, but shows the importance of safety factors! Yikes…

2

Looks like the “challenge” has been around a while… a year at least… and only now it’s been spotted? Does TikTok have the responsibility here for spreading the “challenge” or Amazon for not having better filters? Reminds me of that milk-crate challenge that started up this year and led to serious injuries…

tiktok
tiktok829×534 40 KB

1 Like
3

just logged in to talk about this, here!

How would you deal with an output like this? Or if a product you made suggested such a dangerous thing?

It makes Alexa look pretty naive…I programmed an Alexa skill once called TikTok ideas–a few days later, the notice from Bytedance came :pray:…-- it was basically an Ad Lib–adjectives, verbs, subjects etc… were swapped out each command. any combination of possibilities had a very low risk for inappropriate, or dangerous output. “Go to the mall, buy the biggest pretzel possible and give out compliments to strangers.”

It seems like a flaw in the systems design of the product. If you are drawing information from the environment to affect the algorithm or function, there is an unavoidable risk of dangerous input; because the world inherently has danger.

Using a Sensor in an uncontrolled environment…then using the Sensor data to affect product in real time, creates the potential for this kind of event. How would a computer know that pennies + electricity + trending topic = increased risk of death?