To add to that… when we start looking at the price of things, software engineering becomes very important. Sure I may be able to get the LLM to do something instead of doing it myself, but if it’s costing tens of thousands of dollars to save me one hour of work, it’s going to be hard to justify that cost.

We’ll eventually need to become familiar with which problems it’s relatively cheap to use an LLM for and which problems it will be very expensive to use it for. Hmmmm… estimating performance of an algorithm and making decisions about when to use it and not use it… what does that sound like?