Post

Going to be a bit crude here but the disconnect between management and ground-level folks when it comes to the conversation about (effective) use of GenAI might vanish if the latter were allowed to call out the former for being stupid fucks by actually calling them "stupid fucks". Might help a bit.

In my (limited) experience, ever since GenAI coding tools have become mainstream the conversation about long-term maintenance has gone right out of the window and it's all become a discussion about generating the right solution or prompting the model right or figuring out the right context...

which isn't new, I guess? Some things have always been a certain kind of way in the vast majority of workplaces, a baseline level of suck, I get it. But also, maybe when folks in a move-fast-break-things workplace take GenAI as a tool to move even faster, maybe calling them stupid fucks is the way.

"code is just a means to an end" yes "write code but be prepared to throw it away" very true "we now have tools that write code faster than you can comprehend (or read), said tools have more memory than you but less than they need for whole tasks — we think this'll make you faster" wait what?

I mean, I'm not trying to be an ass here for the fun of it. (What fun?) It's just that everyone who seems to proselytize GenAI tools in earnestness (earnestness as in only preaching what they practice) is in some weird little clique where they don't work in the sort of field most programmers do.

Anyone from a third world country who's done some basic reading in high school knows that statements like "90% of code will be written by AI in a few years" was wished upon a shriveled monkey's paw — I don't understand why so many programmers with online clout don't seem to think that way

I've had a few friends and former colleagues who worked on outsourced projects that came to them via consultancy work. I'm not naming names or being more specific here because it's pointless. Everyone knows what role India plays in the software world, for better or for worse. It's grunt work. But...

It's necessary work. And the people who work on these systems, with modern tools, modern languages and frameworks that were created to make programming more accessible, I don't think they're trained/motivated to approach GenAI tools with the caution+rigor that celebrity programmer of the day might.

0 reposts1 like

To add here though, I think I understand some of why it's going this way. The dream is that GenAI trivializes software development (and maybe most development/processes) that have been outsourced, right? And not speaking purely geographically here. This applies to all sorts of large organizations.

Which, I mean, good luck? If these systems, behemoths that they are, get wrangled by GenAI in a way that minimizes the burden of building and validating and running them mostly, if not entirely? Great! But I'm not sure if the economics of it work out (unless guillotines come into the picture).