What's your product? You kind of need to provide that context - if your team is building some CRUD or calculator app that's a little different to a global engineering team working on products at massive scale.
> don't see a bright future for this profession
Then I'd apply the same logic to every STEM field. I'd rather have a CS degree and understand the tools that are supposed to be building the future.
I don't see a future where everything is AI yet people stop being educated in how AI and software works?
I did Mechanical Engineering years ago and we've had programs to do FEA and CFD analysis for a while now - yet we still have to learn how to do those by hand.
Put it this way - I don't think people who understand AI and software are going to be worse off in the future than those who don't.
I don't think it's relevant here. The point is that there were several work streams that kept several engineers busy full time that now are done by AI almost with little supervision. We do the same amount of work with less people and the software we ship is better on all relevant metrics. If the demand for software stays the same, we'll need less people for the same output.
> I'd rather have a CS degree and understand the tools that are supposed to be building the future.
Sure, we'll always need some experts. But it'll be a riskier career compared to what it was.
I think what you call supervision, I call care. Yes a company could always not care and ship stuff but you’re just kind of trundling towards failure.
I believe in modern parlance some people call it “slop” but that’s a bit of a flattening of what happens. Again, maybe you have a simpler product and doesn’t need a level of care across all dimensions.
It’s just not true for us, AI has made us more ambitious but also has rapidly upsized how much each member has to care.
It so is.
This could be another variation of "I don't need software engineers anymore for my business!" (it's a vibe-coded e-commerce site).
Went from a groups of team of 40 -> 20 and experienced the exact same thing. This was long before AI and agentic workflows though.
It was more a symptom of how much busy work / non-work was required to get anything done.