Understand the Problem

One of the biggest problems I’ve seen descend on engineers recently is the inability to understand the core problem and actually guide an LLM.

With AI it can be incredibly tempting to just throw an issue to an agent and let it figure out the rest. I wrote about this in my /ai page: LLMs are just vehicles. If you don’t know where you’re going or why, then what are you even doing? What happens when the output completely misses the mark because you never took the time to understand the problem, it’s context, and ask clarifying questions? (Heck even Claude knows to do that)

Overall it’s really disappointing to see. I feel like these engineers had a time when they were true problem solvers and made moves back in the day, but now they’re just a copilot to an LLM. Ironic that it was supposed to be the other way around.