AI in Reverse – “The vision of an entire world laid bare for machines to see.”
I’ve gone down the rabbit hole of taking pictures of plants with my iPhone and having it tell me what kind they are. Built right into the camera app, take a picture, swipe up, see if Siri can come up with the name of the plant. It’s greatly expanded my knowledge of plants from next to zero to telling the difference between scarlet firethorn and silverleaf cotoneaster (firethorn has-- you might have guessed-- thorns).
I also happen to have just finished reading Dr. Fei-Fei Li's The Worlds I See. If you’re wondering about artificial intelligence, where it came from, how it works, that book is a great place to start. Turns out AI recognizes scarlet firethorn because someone taught it how to recognize scarlet firethorn by showing it lots and lots of pictures of scarlet firethorn.*
You may recognize Dr. Li’s name from Time’s 2023 list of the 100 Most Influential People in AI, her role as Director of Stanford University’s Human-Centered AI Institute, her former position as Chief Scientist of AI at Google Cloud, or as the driving force behind ImageNet, which "supercharged the development of AI image-recognition systems," according to the Time write up.
ImageNet is a database of over 14 million images divided into more than 20,000 categories, hand-annotated to indicate what objects are pictured. “The vision of an entire world laid bare for machines to see,” as Dr. Li puts it in the book. ImageNet debuted in 2009, and scientists have been using it to train AI algorithms ever since. Machines are taught like children-- this is a flower, this is a bug-- hence the term "machine learning." That’s why I started thinking of it as AI in reverse, because before AI could teach us anything, we had to teach it first.
*If you’re looking for a name for a fantasy warrior, Scarlet Firethorn is not a bad place to start.