@tante Saying AI made you more productive is nothing other than a huge self-own.

@tante

They're more productive if they get someone else to do the checking invisibly and then they take credit. This used to be called secretarial work and it was what women did before women started sometimes to get paid for some of the value of what they did.

@tante

They're more productive if they get someone else to do the checking invisibly and then they take credit. This used to be called secretarial work and it was what women did before women started sometimes to get paid for some of the value of what they did.

@tante I think the secret is simply not to give a damn whether the code does what it's supposed to do or is efficient, and to hope that no one else notices. You'd be surprised how often that works.

@tante We've gone back to the LOC metric (lines of code) for productivity but for anything now. It was wrong in the 80s, even more wrong now.

@tante @maaikees They don’t check anything!! They copypasta code from Bullshit Machine into coding app, run program. Copy errors into Bullshit Machine, copy “solution” into coding app, repeat until no errors, ship.

@tante They are not checking anything and wouldn’t even know what to check for.

@tante I saw one person use lines of code as productivity indicator. I strongly believe we all know that this is no good metric for productivity. So it is hard for me to believe these claims.

@tante I find it believable to have a tenfold increase in productivity – when their primary job is to create misinformation

@tante All they aspire to is the GitHub's Green Wall of Contributions.

@tante

Um they bad at math?

@tante it's definitely a colossal self-own.

There is also the fact that if this made people even twice as productive or less, we'd be seeing the effects already, but there is little evidence that any industry other than AI is really impacted by it.

@tante have tried to generate a very basic essay using the perplexity paid version during very desperate times but learned soon that it really cannot do basic argumentation — even when spoonfed info. I think I wasted valuable time trying to “prompt engineer” lol. But you live and you learn!

@tante I saw that post because one of the people claiming 10x is a former employer. And yes, they didn't program before

@tante So far I’m noticing that people take real stories of increased productivity and replace the actual reasons with word “AI”.

They are as efficient as someone not doing actual work can be. Division by zero is a magical thing.

@tante people who believe that writing more code === faster are obviously either on a junior level, terribly bad developers or no developers at all.

@tante
My team lead likes AI and makes everything with it. Like analysing log files (python "agents") or writing small scripts, asking AI for everything. And he believes he is really productive with it. I don't know...maybe for easy one time tasks. But if he evaluates log files once a weak he is not really saving a lot of time here (starting a script is not really taking time). Maybe he is not really saving time at all cause he is prompting for it every single time instead of having a script.

My worst experience so far: IT colleague who prompted a script to replace a static number in file with an other new static number. This replacing is about once a month. Doubt that it is faster (and easier) starting a script then using an editor and replace by hand. Script was for a non tech person.

@tante ich glaube, dass ganz viele Leute, die nicht regelmäßig mit Leuten von außerhalb des Gymnasial/Unibereichs zu tun haben, die Intelligenz ihrer Mitmenschen massiv überschätzen. Und das soll gar kein front gegen weniger intelligente Leute sein 🤗 es ist halt nur so, dass Lernen und Selbermachen für viele Leute anstrengend ist, sie es deshalb vermeiden und daher auch nie besser darin werden.
Und die es daher als massive Erleichertung empfinden, wenn ihnen etwas Vorgekautes geliefert wird.

@tante They don’t check every line. Some aren’t reading the code/text that’s generated at all. Conversely, they don’t want to read other people’s gen text

@tante There was a small study some time ago with two groups of developers, one using AI, the other not. One had the feeling they were quicker, the other were actually quicker. I'll see whether I can find the link.

@tante I bet they skip that testing part, as they always do.