AI helps most for low-value tasks as well. The real valuable problems are the ones that can’t be solved easily, and AI is usually much less help with those problems (e.g., system design, kernel optimisation, making business decisions). I’ve seen many people say how AI helps them complete more low-value tasks in less time, which is great but not as meaningful as other work that AI is not that good at yet.
You have to get quite sophisticated to use AI for most higher-value tasks, and the ROI is much less clear than for just helping you write boilerplate. For example, using AI to help optimise GPU kernels by having it try lots of options autonomously is interesting to me, but not trivial to actually implement. Copilot is not gonna cut it.
You have to get quite sophisticated to use AI for most higher-value tasks, and the ROI is much less clear than for just helping you write boilerplate. For example, using AI to help optimise GPU kernels by having it try lots of options autonomously is interesting to me, but not trivial to actually implement. Copilot is not gonna cut it.