Machine learning and the art of under-paying people to fix your problems
Can Duruk, writing about what things like “machine learning” often mean:
There’s another interesting angle here, and that’s about how so much of technology is not really technology, but just a bunch of people working really hard… I do wonder, often, how much into the common consciousness it has dissipated that much of “AI” and “Machine Learning” tools are many, many thousands of people doing the work that appears to be done by machines.
Benedict Evans has a good handle on this; how ML manifests in real life is not some omniscient, clairvoyant being but rather an ability to summon a million interns at will. You can’t, yet at least, make AI look at a photo and be able to tell you something a human cannot, but you can make it look at a million pictures.
But, I guess, you can’t even do that really, so you call up [Google’s largest content moderation contractor] Accenture, so they spin up a lot of workers.
When the machines can’t fix your moderation failures, send a bunch of under-paid, under-resourced people in to fix your failures.
Sign up to the Kites can't fly newsletter to get a weekly summary of everything on the site (plus some other cool stuff) in your inbox.
I mean, it’s not like you're going to remember to come back here on your own. URLs are hard.