Note — Oct 27, 2019

Neural Nets Are Just People All the Way Down

Seen in → No.100

Source →

Very good read and important to know. On how layer after layer of data is part of the process of training a neural net (all the way back to the 1960s). Unsurprisingly, each of those layers was initially classified by humans and most of said humans weren’t credited. (The essay is an issue of Vicki Boykis’ excellent Normcore Tech newsletter.)

The movements required to fabricate are complex. While sewing you are often pulling or easing the fabric. Pieces of fabric have to be lined up perfectly or panels won’t match, buttons and holes won’t align and even something as simple as a zipper won’t work. […]

But the dataset was, really, created by hundreds of thousands of people manually identifying what the pictures were. To date, more than 14 million images have been labeled by ImageNet, aka by people from around the world looking at images and clicking on buttons for cents. […]

Simply put, every single piece of decision-making in a high-tech neural network initially rests on a human being manually putting something together and making a choice. […]

Our contention is that every layer of a given training set’s architecture is infused with politics.