No, Snapchat’s My AI chatbot hasn’t gained sentience, though it did freak a lot of people out by posting a Story this week.
As you can see in this example, the My AI profile randomly shared this strange video frame as a Story yesterday, then shortly after, the bot stopped responding to people’s questions.
That had many questioning the meaning of the single frame image. Was this a shot of a roof, maybe the roof where the My AI server lives? Was this an encoded message of some sort, designed to communicate with the outside world? Was My AI actually coming to life?
The last question really freaked people out, because many of them have had a fun time teasing and provoking the bot, in an effort to make it say controversial things.
What if it was taking notes, and now, it was coming alive?
But it’s not, don’t stress.
Shortly after the Story was published, Snap confirmed that My AI had suffered a glitch, and that it was now back to normal.
“My AI experienced a temporary outage that’s now resolved. At this time, My AI does not have Stories feature.”
Interesting wording. “At this time” could imply that Snap is looking to add this functionality in future, and maybe this glitch was part of an expanded test that could see My AI sharing Stories updates, maybe even unique Stories based on your queries to the bot.
Though given the responses, Snap will want to tread carefully on this front, as AI paranoia could trigger some negative reactions, which could do the opposite of endearing people to the artificial character.
Snap’s already experienced several issues with My AI, with users upset that it was automatically added to the top of their inbox, while it’s also had to implement safeguards within the chat process, after some of its responses were deemed inappropriate.
I mean, all AI bots have the potential to go off the rails, as they’re based on web-based inputs, many of which are off the rails themselves, and if it’s being trained on random junk, it’s going to pump out at least some random rubbish itself at times. Which is why you can’t put all your faith in such tools, though the risk of sentience, at least at this stage, remains pretty low.
Generative AI tools are based on statistics and equations, so they’re not “learning” so much as they’re iterating based on the most likely combinations and inputs. That’s still impressive, especially considering the results that they can produce, but it’s not like they’re “thinking”. They’re just a better version of predictive text, which you can obviously tune to a much higher degree, but the actual outputs are of no concern, as such, for the system itself.
So My AI isn’t showing you where it lives, or where you live, or anything else, and you’re not going to wake up and see that weird purple character eyeing you from down the darkened hallway.
Probably. Not yet, at least.