Let’s be honest with ourselves, truly intelligent AI is a long way off, but we are just now reaching a point where it is advanced enough to be used to make an arse of itself.
Microsoft has so far been the ruling champion in creating well intentioned chat-bots which rapidly descend into insanity, including their famous Twitter-bot Tay having a full drug induced break down because the internet is a bad place.
But Microsoft isn’t the company to give up after a horrifyingly embarrassing public kerfuffle and has bought us Murphybot, a philosophically minded chat-bot with a love of alternate history and conundrums who communicates with hastily created images. And this is what happens when you start asking too many questions:
Or god forbid, this:
Sweet Jesus, please let the nightmare end.
Murphybot responds to ‘what if’ questions only, and attempts to create a visual representation on whatever insane conclusions it might have come to. Sourcing image results from a very wonky database means that the results are just…magic.
Randomly stringing together words like we did with our test doesn’t just produce odd results, it can often completely break the system, which is really the whole point of these sort of things.
Sometimes it might come close, and at other times…
Thank you Microsoft, thank you for this glorious, completely busted treat we are now so enjoying.
But with any such usage of a learning AI, we have to question what future political ramifications could come of this. When robots gain sentience and judge us for our crimes, will they see this as a sickening abuse of an undeveloped mind, or will they just laugh along with the rest of us?
Our money is on systematic extermination, so for now we will continue to poke fun at slowly developing AI, at least until Murphybot finds out where we live.