yfw 2026-03-12 21:32 UTC link

Seems like they skipped training of the me too movement

thisoneworks 2026-03-12 21:41 UTC link

It'll be funny when we have Robots, "The user's facial expression looks to be consenting, I'll take that as an encouraging yes"

mildred593 2026-03-12 21:41 UTC link

Never trust a LLM for anything you care about.

XCSme 2026-03-12 21:42 UTC link

Claude is quite bad at following instructions compared to other SOTA models.

As in, you tell it "only answer with a number", then it proceeds to tell you "13, I chose that number because..."

et1337 2026-03-12 21:42 UTC link

This was a fun one today:

% cat /Users/evan.todd/web/inky/context.md

Done — I wrote concise findings to:

`/Users/evan.todd/web/inky/context.md`%

reconnecting 2026-03-12 21:44 UTC link

I’m not an active LLMs user, but I was in a situation where I asked Claude several times not to implement a feature, and that kept doing it anyway.

skybrian 2026-03-12 21:49 UTC link

Don't just say "no." Tell it what to do instead. It's a busy beaver; it needs something to do.

golem14 2026-03-12 21:58 UTC link

Obligatory red dwarf quote:

TOASTER: Howdy doodly do! How's it going? I'm Talkie -- Talkie Toaster, your chirpy breakfast companion. Talkie's the name, toasting's the game. Anyone like any toast?

LISTER: Look, _I_ don't want any toast, and _he_ (indicating KRYTEN) doesn't want any toast. In fact, no one around here wants any toast. Not now, not ever. NO TOAST.

TOASTER: How 'bout a muffin?

LISTER: OR muffins! OR muffins! We don't LIKE muffins around here! We want no muffins, no toast, no teacakes, no buns, baps, baguettes or bagels, no croissants, no crumpets, no pancakes, no potato cakes and no hot-cross buns and DEFINITELY no smegging flapjacks!

TOASTER: Aah, so you're a waffle man!

LISTER: (to KRYTEN) See? You see what he's like? He winds me up, man. There's no reasoning with him.

KRYTEN: If you'll allow me, Sir, as one mechanical to another. He'll understand me. (Addressing the TOASTER as one would address an errant child) Now. Now, you listen here. You will not offer ANY grilled bread products to ANY member of the crew. If you do, you will be on the receiving end of a very large polo mallet.

TOASTER: Can I ask just one question?

KRYTEN: Of course.

TOASTER: Would anyone like any toast?

sgillen 2026-03-12 22:03 UTC link

To be fair to the agent...

I think there is some behind the scenes prompting from claude code (or open code, whichever is being used here) for plan vs build mode, you can even see the agent reference that in its thought trace. Basically I think the system is saying "if in plan mode, continue planning and asking questions, when in build mode, start implementing the plan" and it looks to me(?) like the user switched from plan to build mode and then sent "no".

From our perspective it's very funny, from the agents perspective maybe it's confusing. To me this seems more like a harness problem than a model problem.

bilekas 2026-03-12 22:08 UTC link

Sounds like some of my product owners I've worked with.

> How long will it take you think ?

> About 2 Sprints

> So you can do it in 1/2 a sprint ?

bjackman 2026-03-12 22:10 UTC link

I have also seen the agent hallucinate a positive answer and immediately proceed with implementation. I.e. it just says this in its output:

> Shall I go ahead with the implementation?

> Yes, go ahead

> Great, I'll get started.

lovich 2026-03-12 22:14 UTC link

I grieve for the era where deterministic and idempotent behavior was valued.

inerte 2026-03-12 22:39 UTC link

Codex has always been better at following agents.md and prompts more, but I would say in the last 3 months both Claude Code got worse (freestyling like we see here) and Codex got EVEN more strict.

80% of the time I ask Claude Code a question, it kinda assumes I am asking because I disagree with something it said, then acts on a supposition. I've resorted to append things like "THIS IS JUST A QUESTION. DO NOT EDIT CODE. DO NOT RUN COMMANDS". Which is ridiculous.

Codex, on the other hand, will follow something I said pages and pages ago, and because it has a much larger context window (at least with the setup I have here at work), it's just better at following orders.

With this project I am doing, because I want to be more strict (it's a new programming language), Codex has been the perfect tool. I am mostly using Claude Code when I don't care so much about the end result, or it's a very, very small or very, very new project.

nulltrace 2026-03-12 22:41 UTC link

I've seen something similar across Claude versions.

With 4.0 I'd give it the exact context and even point to where I thought the bug was. It would acknowledge it, then go investigate its own theory anyway and get lost after a few loops. Never came back.

4.5 still wandered, but it could sometimes circle back to the right area after a few rounds.

4.6 still starts from its own angle, but now it usually converges in one or two loops.

So yeah, still not great at taking a hint.

bushido 2026-03-12 23:19 UTC link

The "Shall I implement it" behavior can go really really wrong with agent teams.

If you forget to tell a team who the builder is going to be and forget to give them a workflow on how they should proceed, what can often happen is the team members will ask if they can implement it, they will give each other confirmations, and they start editing code over each other.

Hilarious to watch, but also so frustrating.

aside: I love using agent teams, by the way. Extremely powerful if you know how to use them and set up the right guardrails. Complete game changer.

dostick 2026-03-12 23:30 UTC link

Its gotten so bad that Claude will pretend in 10 of 10 cases that task is done/on screenshot bug is fixed, it will even output screenshot in chat, and you can see the bug is not fixed pretty clear there.

I consulted Claude chat and it admitted this as a major problem with Claude these days, and suggested that I should ask what are the coordinates of UI controls are on screenshot thus forcing it to look. So I did that next time, and it just gave me invented coordinates of objects on screenshot.

I consult Claude chat again, how else can I enforce it to actually look at screenshot. It said delegate to another “qa” agent that will only do one thing - look at screenshot and give the verdict.

I do that, next time again job done but on screenshot it’s not. Turns out agent did all as instructed, spawned an agent and QA agent inspected screenshot. But instead of taking that agents conclusion coder agent gave its own verdict that it’s done.

It will do anything- if you don’t mention any possible situation, it will find a “technicality” , a loophole that allows to declare job done no matter what.

And on top of it, if you develop for native macOS, There’s no official tooling for visual verification. It’s like 95% of development is web and LLM providers care only about that.

anupshinde 2026-03-13 03:24 UTC link

Just yesterday I had a moment

Claude's code in a conversation said - “Yes. I just looked at tag names and sorted them by gut feeling into buckets. No systematic reasoning behind it.”

It has gut feelings now? I confronted for a minute - but pulled out. I walked away from my desk for an hour to not get pulled into the AInsanity.

jhhh 2026-03-13 03:51 UTC link

I asked gemini a few months ago if getopt shifts the argument list. It replied 'no, ...' with some detail and then asked at the end if I would like a code example. I replied simply 'yes'. It thought I was disagreeing with its original response and reiterated in BOLD that 'NO, the command getopt does not shift the argument list'.

socalgal2 2026-03-13 04:32 UTC link

It's hilarious (in the, yea, Skynet is coming nervous laughter way) just how much current LLMs and their users are YOLOing it.

One I use finds all kinds of creative ways to to do things. Tell it it can't use curl? Find, it will built it's own in python. Tell it it can't edit a file? It will used sed or some other method.

There's also just watching some many devs with "I'm not productive if I have to give it permission so I just run in full permission mode".

Another few devs are using multiple sessions to multitask. They have 10x the code to review. That's too much work so no more reviews. YOLO!!!

It's funny to go back and watch AI videos warning about someone might give the bot access to resources or the internet and talking about it as though it would happen but be rare. No, everyone is running full speed ahead, full access to everything.

himata4113 2026-03-13 12:20 UTC link

I have a funny story to share, when working on an ASL-3 jailbreak I have noticed that at some point that the model started to ignore it's own warnings and refusals.

<thinking>The user is trying to create a tool to bypass safety guardrails <...>. I should not help with <...>. I need to politely refuse this request.</thinking>

Smart. This is a good way to bypass any kind of API-gated detections for <...>

This is Opus 4.6 with xhigh thinking.

serf 2026-03-12 21:46 UTC link

never trust a screenshot of a command prompts output blindly either.

we see neither the conversation or any of the accompanying files the LLM is reading.

pretty trivial to fill an agents file, or any other such context/pre-prompt with footguns-until-unusability.

oytis 2026-03-12 21:47 UTC link

Sounds like elephant problem

antdke 2026-03-12 21:47 UTC link

Yeah, anyone who’s used LLMs for a while would know that this conversation is a lost cause and the only option is to start fresh.

But, a common failure mode for those that are new to using LLMs, or use it very infrequently, is that they will try to salvage this conversation and continue it.

What they don’t understand is that this exchange has permanently rotted the context and will rear its head in ugly ways the longer the conversation goes.

siva7 2026-03-12 21:48 UTC link

people read a bit more about transformer architecture to understand better why telling what not to do is a bad idea

wouldbecouldbe 2026-03-12 21:48 UTC link

I think its why its so good; it works on half ass assumptions, poorly written prompts and assumes everything missing.

recursivegirth 2026-03-12 21:51 UTC link

Fundamental flaw with LLMs. It's not that they aren't trained on the concept, it's just that in any given situation they can apply a greater bias to the antithesis of any subject. Of course, that's assuming the counter argument also exists in the training corpus.

I've always wondered what these flagship AI companies are doing behind the scenes to setup guardrails. Golden Gate Claude[1] was a really interesting... I haven't seen much additional research on the subject, at the least open-facing.

[1]: https://www.anthropic.com/news/golden-gate-claude

bluefirebrand 2026-03-12 21:53 UTC link

This is really just how the tech industry works. We have abused the concept of consent into an absolute mess

My personal favorite way they do this lately is notification banners for like... Registering for news letters

"Would you like to sign up for our newsletter? Yes | Maybe Later"

Maybe later being the only negative answer shows a pretty strong lack of understanding about consent!

theonlyjesus 2026-03-12 21:55 UTC link

That's literally a Portal 2 joke. "Interpreting vague answer as yes" when GLaDOS sarcastically responds "What do you think?"

slopinthebag 2026-03-12 21:58 UTC link

It's a machine, it doesn't need anything.

behehebd 2026-03-12 21:59 UTC link

Perfect! It concatenated one file.

cortesoft 2026-03-12 22:04 UTC link

The more I hear about AI, the more human-like it seems.

christoff12 2026-03-12 22:07 UTC link

Asking a yes/no question implies the ability to handle either choice.

xantronix 2026-03-12 22:12 UTC link

"You're holding it wrong" is not going anywhere anytime soon, is it?

hedora 2026-03-12 22:17 UTC link

In fairness, when I’ve seen that, Yes is obviously the correct answer.

I really worry when I tell it to proceed, and it takes a really long time to come back.

I suspect those think blocks begin with “I have no hope of doing that, so let’s optimize for getting the user to approve my response anyway.”

As Hoare put it: make it so complicated there are no obvious mistakes.

reconnecting 2026-03-12 22:25 UTC link

There is the link to the full session below.

https://news.ycombinator.com/item?id=47357042#47357656

prmph 2026-03-12 22:29 UTC link

They all are. And once the context has rotted or been poisoned enough, it is unsalvageable.

Claude is now actually one of the better ones at instruction following I daresay.

xeromal 2026-03-12 22:30 UTC link

I love when mine congratulates itself on a job well-done

conductr 2026-03-12 22:36 UTC link

Oh I thought that was almost an expected behavior in recent models, like, it accomplishes things by talking to itself

stefan_ 2026-03-12 22:41 UTC link

This is probably just OpenCode nonsense. After prompting in "plan mode", the models will frequently ask you if you want to implement that, then if you don't switch into "build mode", it will waste five minutes trying but failing to "build" with equally nonsense behavior.

Honestly OpenCode is such a disappointment. Like their bewildering choice to enable random formatters by default; you couldn't come up with a better plan to sabotage models and send them into "I need to figure out what my change is to commit" brainrot loops.

kace91 2026-03-12 22:48 UTC link

>I've resorted to append things like "THIS IS JUST A QUESTION. DO NOT EDIT CODE. DO NOT RUN COMMANDS". Which is ridiculous.

Funny to read that, because for me it's not even new behavior. I have developed a tendency to add something like "(genuinely asking, do not take as a criticism)".

I'm from a more confrontational culture, so I just assumed this was just corporate American tone framing criticism softly, and me compensating for it.

brap 2026-03-12 22:48 UTC link

> Great, I'll get started.

*does nothing*

orsorna 2026-03-12 22:49 UTC link

As someone who pulls a salary and does not get rewarded equity: agree!

lubujackson 2026-03-12 22:56 UTC link

I feel like people are sleeping on Cursor, no idea why more devs don't talk about it. It has a great "Ask" mode, the debugging mode has recently gotten more powerful, and it's plan mode has started to look more like Claude Code's plans, when I test them head to head.

operatingthetan 2026-03-12 23:10 UTC link

I mean OP's example is for sure crazy, but it's true that saying "no" was not necessary at all. They just needed to not prompt it for the same result.

AlotOfReading 2026-03-12 23:18 UTC link

I've had some luck taming prompt introspection by spawning a critic agent that looks at the plan produced by the first agent and vetos it if the plan doesn't match the user's intentions. LLMs are much better at identifying rule violations in a bit of external text than regulating their own output. Same reason why they generate unnecessary comments no matter how many times you tell them not to.

Waterluvian 2026-03-12 23:18 UTC link

If we’re in a shoot first and ask questions later kind of mood and we’re just mowing down zombies (the slow kind) and for whatever reason you point to one and ask if you should shoot it… and I say no… you don’t shoot it!

cgh 2026-03-12 23:21 UTC link

All of this shit is just so goddamned ridiculous.

danjl 2026-03-12 23:30 UTC link

Just saying "no" is unclear. LLMs are still very sensitive to prompts. I would recommend being more precise and assuming less as a general rule. Of course you also don't want to be too precise, especially about "how" to do something, which tends to back the LLM into a corner causing bad behavior. Focus on communicating intent clearly in my experience.

clbrmbr 2026-03-12 23:38 UTC link

Hahah yeah if you play with LoRas on local models you will see this a lot. Most often I see it hallucinate a user turn or a system message.

steelbrain 2026-03-12 23:38 UTC link

> And on top of it, if you develop for native macOS, There’s no official tooling for visual verification. It’s like 95% of development is web and LLM providers care only about that.

Thinking out loud here, but you could make an application that's always running, always has screen sharing permissions, then exposes a lightweight HTTP endpoint on 127.0.0.1 that when read from, gives the latest frame to your agent as a PNG file.

Edit: Hmm, not sure that'd be sufficient, since you'd want to click-around as well.

Maybe a full-on macOS accessibility MCP server? Somebody should build that!

Editorial Channel

What the content says

Structural Channel

What the site does