AI use damages professional reputation, study suggests
arstechnica.com
arstechnica.com
I’ve been designing and developing software for 40 years. In my experience, those who resist new technology often find themselves left behind. Those who embrace it, genuinely learning how to apply it rather than just crafting feel-good narratives for leadership, tend to thrive.
My experience so far with AI and code design and generation is it's not a magic bullet. Like every tool that's come before that was supposedly going to replace developers and software engineers or dramatically reduce the need for developers and software engineers, it's not. It's another tool for developers and software engineers to use to get even more work done. Paradoxically, as more work gets done the backlog increases with even more work to do. It's been this way for the 40 years I've been in this industry, and the "old timers" when I started indicated it had been that way throughout their career, too.
I haven’t professionally built anything on bare metal in years. It sits somewhere in the aether these days, and I no longer have to worry about it in the same way.
I can’t speak for other languages, but Spring Boot made enterprise level Java development accessible to the masses, and the jobs are still there. They have just shifted.
If you haven’t done the groundwork and seen what these frameworks are replacing, you won’t truly grasp what the tools are doing. That is why vibe coding leads to unintended side effects, and I feel for students who are dropped straight into frameworks that obscure what is going on under the bonnet. It is not magic, it just looks like it.
If an organisation treats AI as a silver bullet, the best they will get is an off-the-shelf assistant. The real gains will come from domain-specific models that understand the business well enough to generate tailored, useful output. But that does not remove the engineer, it just shifts their role towards quality, safety, and accountability.
Compliance alone – GDPR, PCI, and the rest – will slow down any industry-wide adoption of truly bespoke AI models. Retreading the same ground just to offer suggestions that feel human? That could cost tens, if not hundreds, of millions per organisation.