A year ago, ChatGPT appeared and everyone seemed to have an opinion about it. Some people called it the end of programming. Others called it a toy. As an engineer, I had only one reasonable reaction: find out how it works.
Studying before judging
In September 2023 I completed courses in prompt engineering, artificial intelligence and machine learning. The goal was not to collect certificates. I wanted a vocabulary that let me separate what these systems really do from what the headlines say they do.
A prompt is a specification
The most useful idea I took away is simple. A good prompt looks a lot like a good technical specification: it gives context, states constraints, and describes what a correct result looks like. There is nothing mystical about it. Engineers who have written tickets, runbooks and network diagrams already have most of the skill.
What I am watching
The same model can sound brilliant and be completely wrong in the same paragraph. That means the person using it still has to know the subject. For now, I see these tools as a very fast assistant whose work must always be reviewed.
Every new technology in my career has followed the same pattern: first the noise, then the learning, and finally the quiet moment when it becomes part of the job. I intend to be ready for that moment.