argumate:

bubobubosibericus:

catleidh:

lierdumoa:

What we all need to understand about AI in a nutshell:

There’s an algorithm that can reliably predict, from aggregate facebook posts, the onset of a manic episode in a person suffering from bipolar disorder – more reliably even, than a trained psychotherapist, who only has access to the information a patient provides them in therapy sessions. 

“Won’t technology like that help people with bipolar disorder?”

Theoretically, it could. But this algorithm wasn’t designed to help people with bipolar disorder.

This algorithm was designed to sell plane tickets to Las Vegas.

[source]

Guys you gotta click the link. Because this was not just an accidental outcome- it’s click bait based on algorithms.

the AI knows you’re about to hit mania so it pushes Las Vegas Airfare Adverts to you so when you hit the mania you buy one, because it’s figured out that bipolar people that have hit mania are more likely to start gambling/making impulsive decisions – and that will sell a ticket. 

They didn’t explicitly mean to target bipolar people who have hit mania, but the algorithm is really good at learning shit and the more things you like or sites you visit the more accurate it’s recommendations are, so it basically started pushing ads at people with bipolar disorder and nobody knows what exactly it’s picking up on.

Seriously click the source link it’s a TED Talk about 23 minutes long.

@argumate relevant.

arguably as with most AI risk it’s just the intensification of an existing trend, in this case “advertising takes advantage of people for profit”