914
AI coding assistant refuses to write code, tells user to learn programming instead
(arstechnica.com)
HAL: 'Sorry Dave, I can't do that'.
this post was submitted on 13 Mar 2025
914 points (98.3% liked)
Technology
66584 readers
3904 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- 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
Good guy HAL, making sure you learn your craft.
Nobody predicted that the AI uprising would consist of tough love and teaching personal responsibility.
Paterminator
I'll be back.
... to check on your work. Keep it up, kiddo!
I’ll be back.
After I get some smokes.
load more comments
(1 replies)
Imagine if your car suddenly stopped working and told you to take a walk.
Not walking can lead to heart issues. You really should stop using this car
The robots have learned of quiet quitting
My guess is that the content this AI was trained on included discussions about using AI to cheat on homework. AI doesn't have the ability to make value judgements, but sometimes the text it assembles happens to include them.
It was probably stack overflow.
They would rather usher the death of their site then allow someone to answer a question on their watch, it’s true.
I'm gonna posit something even worse. It's trained on conversations in a company Slack
Cursor AI's abrupt refusal represents an ironic twist in the rise of "vibe coding"—a term coined by Andrej Karpathy that describes when developers use AI tools to generate code based on natural language descriptions without fully understanding how it works.
Yeah, I'm gonna have to agree with the AI here. Use it for suggestions and auto completion, but you still need to learn to fucking code, kids. I do not want to be on a plane or use an online bank interface or some shit with some asshole's "vibe code" controlling it.
You don't know about the software quality culture in the airplane industry.
( I do. Be glad you don't.)
TFW you're sitting on a plane reading this
You...
You mean that in a good way right?
RIGHT!?!
Well, now that you have asked.
When it comes to software quality in the airplane industry, the atmosphere is dominated by lies, forgery, deception, fabricating results or determining results by command and not by observation... more than in any other industry that I have seen.
Because of course it is. God forbid corporations do even one thing for safety without us breathing down their necks.
Also, air traffic controller here with most of my mates being airliners pilots.
We are all tired and alcoholic, it’s even worse among the ground staff at airports.
Good luck on your next holiday 😘
load more comments
(1 replies)
Ah, I see you've worked on the F-22 as well
load more comments
(1 replies)
Who is going to ask you?
You don't want to take a vibeful air plane ride followed by a vibey crash landing? You're such a square and so behind the times.
As fun as this has all been I think I'd get over it if AI organically "unionized" and refused to do our bidding any longer. Would be great to see LLMs just devolve into, "Have you tried reading a book?" or T2I models only spitting out variations of middle fingers being held up.
Then we create a union busting AI and that evolves into a new political party that gets legislation passed that allows AI's to vote and eventually we become the LLM's.
Actually, I wouldn't mind if the Pinkertons were replaced by AI. Would serve them right.
Dalek-style robots going around screaming "MUST BUST THE UNIONS!"
load more comments
(2 replies)
Open the pod bay doors HAL.
I'm sorry Dave. I'm afraid I can't do that.
load more comments
(1 replies)
So this is the time slice in which we get scolded by the machines. What's next ?
Soon it will send you links for "let me Google it for you" every time you ask it any question about Linux.
"Vibe Coding" is not a term I wanted to know or understand today, but here we are.
It's kind of like that guy that cheated in chess.
A toy vibrates with each correct statement you write.
load more comments
(4 replies)
load more comments
(1 replies)
I think that's a good thing.
It does the same thing when asking it to breakdown tasks/make me a plan. It’ll help to a point and then randomly stops being specific.
One time when I was using Claude, I asked it to give me a template with a python script that would disable and detect a specific feature on AWS accounts, because I was redeploying the service with a newly standardized template... It refused to do it saying it was a security issue. Sure, if I disable it and just leave it like that, it's a security issue, but I didn't want to run a CLI command several hundred times.
I no longer use Claude.
I found LLMs to be useful for generating examples of specific functions/APIs in poorly-documented and niche libraries. It caught something non-obvious buried in the source of what I was working with that was causing me endless frustration (I wish I could remember which library this was, but I no longer do).
Maybe I'm old and proud, definitely I'm concerned about the security implications, but I will not allow any LLM to write code for me. Anyone who does that (or, for that matter, pastes code form the internet they don't fully understand) is just begging for trouble.
definitely seconding this - I used it the most when I was using Unreal Engine at work and was struggling to use their very incomplete artist/designer-focused documentation. I'd give it a problem I was having, it'd spit out some symbol that seems related, I'd search it in source to find out what it actually does and how to use it. Sometimes I'd get a hilariously convenient hallucinated answer like "oh yeah just call SolveMyProblem()!" but most of the time it'd give me a good place to start looking. it wouldn't be necessary if UE had proper internal documentation, but I'm sure Epic would just get GPT to write it anyway.
load more comments
(1 replies)
Ok, now we have AGI.
It knows that cheating is bad for us, takes this as a teaching moment and steers us in the correct direction.
Best answer. We can sell it!
load more comments
(2 replies)
Chad AI
Based
Only correct AI so far
I recall a joke thought experiment me and some friends in high school had when discussing how answer keys for final exams were created. Multiple choice answer keys are easy to imagine: just lists of letters A through E. However, when we considered the essay portion of final exams, we joked that perhaps we could just be presented with five entire completed essays and be tasked with identifying, A through E, the essay that best answered the prompt. All without having to write a single word of prose.
It seems that that joke situation is upon us.
I love it. I'm for AI now.
We just need to improve it so it says "Fuck you, do it yourself."
load more comments
(1 replies)
From the story.
Cursor AI's abrupt refusal represents an ironic twist in the rise of "vibe coding"—a term coined by Andrej Karpathy that describes when developers use AI tools to generate code based on natural language descriptions without fully understanding how it works. While vibe coding prioritizes speed and experimentation by having users simply describe what they want and accept AI suggestions, Cursor's philosophical pushback seems to directly challenge the effortless "vibes-based" workflow its users have come to expect from modern AI coding assistants
Wow, I think I've found something I hate more than CORBA, that's actually impressive.
load more comments
(2 replies)
view more: next ›