On my journey to catch back up with ML after having abandoned it around 2000, I find myself thinking the same things as Karpathy. There is something profound h

On my journey to catch back up with ML after having abandoned it around 2000, I find myself thinking the same things as Karpathy. There is something profound here, and as someone that has been writing software for 40 years since the ripe age of 8, it is scintillating in its potential. I'm working in a field distant from most current applications of ML (DDOS detection/mitigation and "cyber" security), and so I'm reaping the benefits of being able to take this stuff and apply it to lots of problems. Results are shocking so far. It feels more like training a cybernetic animal than programming, and I'm ok with that. It's a puzzle of trying to figure out how to morph reality so that ML systems can gain traction, and that's just a fascinating new way of looking at problems. Great article!