back
1 comments
While I consider myself relatively well informed about machine learning's strengths, I keep on being surprised by just how many areas/industries have niches that are perfect fits for ML to be applied. I didn't even know about protein folding before DeepMind solved it, and I would have never expected RL to be applied to chip design.

Tangentially related: my sense is that, in the coming couple of years, ML will significantly lower the cost of software development[0], starting from frontend development[1], and then backend, and so on.

[0] Github Copilot is an undeniable demonstration of this.

[1] Starting from frontend because I think the "implement this Figma mock" task is particularly susceptible to ML. one such startup: https://uizard.io/