Funny enough, I saw my first ad for Alexa's LLM chatbot functionality only just this morning and my first thought was how long it would take before that feature blows up in Amazon's face.
It’s well known that Alexa uses knowledge from “an Alexa answers contributor.” It’s also stated verbally. There have been plenty of answers I’ve received which are incorrect and/or harmful. Many I know who use Alexa have realized this. I personally don’t use Alexa to answer any questions outside of actuating lights and getting package delivery notifications.
Perhaps harmful is too strong. When asking about calories or carbs the answers are wrong. Also if I ask if my dog can have X I can’t trust it. I have to google. For example radishes.
Are we surprised that AIs can't withstand the levels of BS that are thrown their way (either by corporate shills or by high-profile grassroots campaigns)?
Even the traditional media and social media are heavily full of what I'll call nicely: inconsistent and irrational viewpoints.
7 comments
[ 3.0 ms ] story [ 25.8 ms ] threadEven the traditional media and social media are heavily full of what I'll call nicely: inconsistent and irrational viewpoints.