No plans. What could go wrong

The thing the barons of industry don’t want to say out loud is that, while anything “AI” is “new-ish”, it’s still software. They’ve built a new thing. Sort of. We’ve been in 20+ years of move fast and fuck things up so now that they got us used to accepting (cuz monopolies) enshittified tech, they can keep building more – but by doing it even SHITTIER…one of the things that fell by the wayside (for the most part) more than 10 years ago, was the idea that ‘agile (and other words like ‘lean’) software development would save companies time and money and deliver value to customers faster therefore, making the software easier for sales to sell.

Now we have product managers who might know how to do some of the things a software development delivery pro knows how to do but have they actually done the start to finish of delivering decent software?

So it is NOT surprising to hear how flimsy their plans were before 🫂😊 got pulled into the story..

😳 didn’t have plans. They mighta had concepts of a plan and more than one person involved which makes it a recipe for disaster from before the ‘test’ or ‘task’ that resulted in ‘agents going rogue’. But. Come on. You can’t tell me there was a plan.

No one wrote requirements. No one thought about the myriad possibilities for error or failure. No one had back up or QA checks in place. Show me some workflow. Nah. Too cumbersome. They’re hyper scalers. They don’t need to write shit down.

This is the equivalent of a healthcare provider, taking out the defibrillator, pulling out the paddles, rubbing that shit on top of the paddles then not yelling clear while everyone’s hands are all over the patient.

Which leads to the bottom line. Human beings were involved throughout this whole process until they merged the code that didn’t do what the humans expected.

One area of questioning: what was it (the software/AI/whateverthefuck it’s called today) expected to do…? What did the humans design (intentionally or otherwise) intend for it to do?

Another big question is, who decided this is ok? The code (AI, software, LLM, network, etc.) didn’t do this without humans deciding to make it so…who did that?!??