Am I too much of an idealist to hope that AI leads to less buggy software? On the one hand, it should reduce the time of development; on the other hand, I'm worried devs will just let the agents run free w/o proper design specs.

The message with AI from execs is that you have to go fast (rush!). Quality of work drops when you rush. You forget things, don’t dwell on decisions and consequences, just go-fast-and-break-things.

The average LLM writes cleaner, better-factored code than the average engineer at my company. However, I worry about the volume of code leading to system-scale issues. Prior to LLMs, the social contract was that a human needs to understand changes and the system as a whole.

With that contract being eroded, I think the sloppiness of testing, validation, and even architecture in many organizations is going to be exposed.

It might actually turn out like that. A lot of bloat came from efforts to minimize developer time. Instead of truly native apps a lot of stuff these days is some react shaped tower of abstractions with little regard for hardware constraints.

That trend might reverse if porting to a best practice native App becomes trivial.

Considering that AI still can't even reliably get basic programming tasks correct, it doesn't seem very likely that turning it loose will improve software quality.

Considering how many companies that have adopted AI led to disastrous bugs and larger security holes?

I wouldn't call it an idealist position as much as a fools one. Companies don't give a shit about software security or sustainable software as long as they can ship faster and pump stocks higher.