this post was submitted on 14 Mar 2026
627 points (98.5% liked)
Technology
82798 readers
3599 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related news or articles.
- Be excellent to each other!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, this includes using AI responses and summaries. To ask if your bot can be added please contact a mod.
- Check for duplicates before posting, duplicates may be removed
- Accounts 7 days and younger will have their posts automatically removed.
Approved Bots
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
There must be something really seriously wrong at Microsoft. I can understand that Windows patches are complex and that they might break some of those crazy things people are running on their machines. But how is a bug that is killing access to the C:\ drive able to get through testing? WTF are they doing?
It's going to come out that there's AI in the code. And the code testing was done by AI, who gave the buggy code the green light.
Or worse: AI is doing the QA as well
What QA? Microsoft's QA was always the CEO demoing the latest repository head on stage.
They at least used to be embarassed by a live BSOD.
We're doing the QA.
"Code Testing" = QA
It passed the unit test, it must be good!
Quality Artificial Intelligence assurance
They don’t need testing because they tell the ai to not make any errors
my boss loves AI and he uses it for everything. he made some stats graphs and summaries, and he was bragging how he got AI to make them errorless: he tells it to check for errors and makes it swear it's accurate... while we were looking at a graph where the y column numbers were all fucked up
Interestingly, AI is actually pretty good at making graphs, the trick is you don't ask it to actually make the graph itself. Instead you have to ask it to write a python script to create a graph using matplotlib from whatever source file contains the data, then run that script. Same with math. Don't ask it to do math directly, instead ask it to write a bash or python script to do some math, then run that. Still not perfect, but your success rate increases by about 1000%
Because of so much open source and stack overflow it was trained on.
But who writes bash scripts to do math?
A full script? Nobody. But you can just run it interactively on the command line, which a lot of AI clients have access to. bc works great for basic math in the shell.
That's about 90% of what I use AI for right there: silly little bash and python scripts. A graph, some image compression, ffmpeg video shenanigans, the works.
It's really really bad at doing spreadsheet analysis. Even basic shit that I would give to an intern. At least an intern with generally just make shit up and pretend it's not wrong even when I point it out, and if they do I get a new intern.
And then the LLM says something like "You're absolutely right, there was an error in that code that is clear and obvious now it has been pointed out and despite the fact you gave the instruction to make no errors. Is there anything else I can help with?"
... and they'll be too blind to take that as the warning it is and continue to ask even more of the LLM.
It's Microslop. This is what's wrong. Also, that they fired too much of the testing staff in favor of (user-)testing rings.
It's not as bad as that time they permanently deleted user documents and photos.
See they had this trick where if you didn't have enough space on your drive to unpack an update, they'd just move your shit to OneDrive temporarily, then move it back when the update was done. Only they forgot to move it back, and lost it. Oops.
Seriously?!?! 😲
No, but are we gonna wait until they do?
My company is starting to roll out having AI both put up PRs AND give code reviews.
I would not be surprised to hear Microslop is doing the same thing and having horrible results.
Amazing what happens when you try to turn your talent pool into lifeless casino monitors.
Vibecoding. Microslop has peddled AI so much that they have gotten addicted to their own supply.
Probably AI code getting tested by AI.
No one smart is going into windows dev in 2026. It’s like working on IBM mainframes. Only people left to work are middle of the road new grads they hire and boomers who are retiring.
*Microslop
Let's start calling it what it is