When we started at university, neural networks and deep learning (the main underlying techniques most AI products use today) were just too slow and hard to work with. By the time we finished our doctorates, that had changed. Tools like TensorFlow made this stuff easier and fun, and the increase in GPU access made training and inference possible for people without Big Tech budgets. For quite a while, we were genuinely excited about the potential.
Then things started to change.
It’s hard to say exactly when. Maybe it was always like this and we just didn’t see it. But by the end of 2020 (a year famous for absolutely nothing world changing whatsoever happening /s) it was clear that the AI we liked was not what the tech companies were interested in. They were increasingly about generative imagery, chatbots writing your material for you, and summaries of art instead of exposure to it. Efforts to mitigate known problems (reinforcing cultural biases, being difficult to make deterministic or explainable) were disparaged and diminished. Researchers and developers who raised concerns were being fired.
Things have only gotten worse since.
If you look at what AI companies promote now, it’s not what we wanted. When you boil down everything they say and strip it right back, what they make are tools to either fire people or demand more work without hiring anyone new to help. That’s the problem AI companies want to solve.
Anything else they achieve is a happy accident on the road to firing as many of your friends and colleagues as possible.
Read the rest at the Yarn Spinner blog
Add new comment