[analog31]

Most people with "engineer" titles spend relatively little of their time on actual quantitative engineering or "higher level" thinking. A lot of their work involves manual information processing: Organizing and arranging things, fitting things together, troubleshooting. This could be justified for a couple of reasons: Maybe a lot of the stuff that was "engineering" is now handled by the CAD software. That's great. But also, the efficiency of those tools has raised the complexity of systems to the point where the interaction between parts consumes most of the engineers' attention.

Managers also spend most of their time on the same things, but handling different kinds of information.

But CAD hasn't changed the immutable laws of engineering, such as Brooks's Law. When I hear about the wonders of AI transforming engineers into higher level thinkers, my snarky response is: "Does this mean that projects will finish on time?"

[withinboredom]

If your engineers (software or otherwise) aren’t spending a lot of time engineering, then you’ve got a hiring problem. Most jobs I’ve worked as a software engineer are 90% engineering (soft and hard skills) and only about 10% programming. With AI, it becomes about 60% engineering and 20% babysitting an AI, and 30% programming because the AI got it wrong.

Now, we can’t even hand this stuff off to juniors and teach them things they’ll hopefully remember. Instead, I have to explain to an AI, for the 60th time, that it has hallucination problems.

Personally, I’d rather have the juniors back.