Email or username:

Password:

Forgot your password?
Misty

Oh man, this news story. Former MLA from Manitoba asks Meta's own "ask AI" chatbot if a phone number he found on Google is really Meta support. It tells him yes. It isn't, and he's scammed out of $500 and control of his accounts. cbc.ca/news/canada/manitoba/fa

16 comments
cancel

@misty i really appreciate the directness and speed with which the dipshits in charge of these things have driven us toward a cliff

cancel

@misty in a normal sci fi story they'd tiptoe around it a while instead of beelining for fucking everything up

phillmv

@misty

this to me speaks so much about what’s ethically dubious about how the tool has been rolled out.

it’s one thing to present it as a choose your own adventure spicy autocomplete for code. code either works or it doesn’t.

but the way it’s been baked into everything else presents it as _authoritative_. it’s not spicy autocomplete it’s ARTIFICIAL INTELLIGENCE.

you need an incredible amount of… “computer literacy” to use it safely, but then its shoved everywhere.

phillmv

@misty

remember that story a few months ago about how Air Canada tried to say that they weren’t liable for the misinformation that the chat bot on its website provided?

its like execs are super stoked for that outcome *everywhere*, just always being accountability-free at all times. /rant 😝

Misty

@phillmv Putting it directly in Facebook is just deliberately putting it in front of the users who are the least prepared to understand what the tech actually is and how unsafe asking it that question is.

ShadSterling

@misty if we had a civil analog of a public defender they would have a field day with this

we're ALL jordan

@misty I feel like between this and the Air Canada refund thing a few months ago, Canada is really the actionable lawsuits on shitty LLMs

MarkS

@misty

I believe this was the same excuse the lawyer used when the court discovered that all the references in his brief were fabrications. After having the AI write the brief, he asked the AI if all the references were real and it said "yes".

Jay

@markstahl @misty Con Artist Assures Mark Con Isn’t Scam, news at 11.

How do these people function in everyday life?

Ratsnake Games

@misty Has AI misinformation gotten someone killed yet? If not, I feel like it's only a matter of time.

Tim 🎮

@ratsnakegames @misty it's already gleefully identifying absolutely lethal mushrooms as completely safe to eat and tasty, it's almost definitely lead to deaths already. Unfortunately those will be a lot harder to prove, making the connection to some LLM's disinformation will be like that urban legend of an icicle killing someone with a blow to the head and then melting away into nothing

Panama Red

@misty IANAL, but I'm thinking he has a case.

persistentdreamergames

@misty

This is the coveted technology billionaires are blowing money on. They want everything thing automated so they don't have to pay workers, but the tech itself is shit and they're killing their own businesses to do this.

FoolishOwl

@misty From the account in the article, it seems like Gaudreau was being careful and was trying to verify a telephone number from what he reasonably expected would be an authoritative source.

This is absolutely Meta's fault.

Umberto Ecco

@misty @mcc can you imagine if Meta *did* have phone support, though! They’d have to hire like 1% of the earth’s population to staff it…

Actually, that would be a good law to fight these big companies. For every x active users, you must have one (1) CSR

SpaceLifeForm

@misty

Does this mean Meta Support is just an hallucination?

Go Up