I'd argue that part of the value of software is its quality and this has business impact. You can't really prompt "give me the best possible quality in existence" to AI. This is something that engineers have always done and it's technical and still relevant.

I mention this because the article is overlooking this as does the industry at large. There is no distinction between different levels of quality of software and how it impacts the business. It has always been underestimated. Every place I've worked in has had huge problems with constant bugs and regressions anytime something is changed. This comes about because software is seen as either existing or not, which is a false dichotomy. AI is making this worse. Engineers need to remain focused on technical quality especially now that the market is going to be saturated with lower quality code.

AI code generation has revealed the value of software that was always there: trust and judgement.