I was really surprised to see Microsoft release another AI Bot so soon after the massive failure that was Tay, but they did, and it is amusing.
The new AI Bot is called CaptionBot, and the way it works is fairly simple. You upload a photo and it tries to caption it based on what the algorithm sees in the photograph. Microsoft’s Cognitive Team spent a great deal of time and money researching how to model things in photographs so computers can understand what’s going on in the photos.
But, when it comes to yoga, it has absolutely no f-ing idea what is going on in these photos.
Being the easily amused geeky yogi that I am, I decided to upload a series of stock yoga photos (including indoor yoga, outdoor yoga, and aerial yoga) to CaptionBot.
This one isn’t too far off, but I don’t really understand how someone can stand on a mountain lake. Unless they’re channeling Jesus.
She is neither lying down, nor on a bed. But nice try, CaptionBot.
Look at the smile on her face! That must be one sexy frisbee.
Ah, yes, “frisbee” — the term of endearment every woman hopes to hear while getting her yoga on with a banyan tree.
Holding an invisible frisbee while doing reverse prayer is definitely new.
She’s throwing an invisible frisbee. With her eyes.
She’s in for a nasty surprise if he throws her in the air like a frisbee.
They’re jumping so well they actually look like they’re standing still, and just what is it with the frisbee obsession??
They too are jumping so slowly it looks like they haven’t.
If you can do camel pose on a skateboard, you’re probably making a lot of money as a professional stunt doer.
Ah yes, the famous invisible tennis racket…
…which you would likely find in the home of a woman who has an invisible table in a room…
…and holds invisible dogs with her feet.
The problem with throwing rivers is if you don’t throw it just right, the river comes back to you like a boomerang.
CaptionBot grammar strong suit is not.
Yeah, not so much.