Keeping up with the Craft

I am pleasantly surprised by what I saw at UtahJS conference! I admit that I went thinking I held a rare minority opinion on LLMs, but then about each session I attended and each conversation I had seemed to echo similar opinions to mine. I think there is hope for a better future! The struggle is real, and I believe that human experience can come out on top.

What is my opinion on LLM technology? In short, it is that software engineers should still “do the repetitions” to master the craft, and we will need to slow things down in order to make things in high quality code. Code is indeed cheap (granted it was always cheap), but the expertise behind it is not.

And so I heard at UtahJS Conference from several presenters and conversations a similar vibe. There is a displeasure at business directives to create code so cheaply, because the expertise feels disregarded. Also, our expertise comes from doing the work to find new and progressively enhanced features. Using LLMs may atrophy that expertise.

I do realize there is a bias here too. I deliberately chose not to attend sessions that seemed likely to evangelize LLMs. However, there were some sessions I couldn’t attend because of scheduling conflicts. But the point is that I recognize the pro-LLM sentiment could still be present.

And so I recognize the differing opinions and I welcome a discussion on how we can properly take advantage of LLM technology about avoiding blowbacks, or unintended consequences. A lot of scientific work is already showing some unintended consequences. I know LLM technology can be useful, but the evangelizing of it is making real use cases unclear.