Hacker Newsnew | past | comments | ask | show | jobs | submit | antdke's commentslogin

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.


I’ve found this happens with repos over time. Something convinces it that implementing the same bug over and over is a natural next step.

I’ve found keeping one session open and giving progressively less polite feedback when it makes that mistake it sometimes bumps it out of the local maxima.

Clearing the session doesn’t work because the poison fruit lives in the git checkout, not the session context.


I like how anything these tools do wrong just boils down to “you’re using it wrong”

It can do no wrong

It is unfalsifiable as a tool


I don't think it's intended as that kind of binary. It's more like "yeah, it's flawed in that way, and here's how you can get around that". If someone's claiming the tool is perfect, they're wrong; but if someone's repeatedly using it in the way that doesn't work and claiming the tool is useless, they're also wrong.


Nobody said that. But as you say, it's just a tool. Tools need to be used correctly. If tools are unintuitive, maybe that's due to the nature of the tool or due to a flaw in it's design. But either way, you as the user need to work around that if you want to get the maximum use out of the tool.


Well, imagine this was controlling a weapon.

“Should I eliminate the target?”

“no”

“Got it! Taking aim and firing now.”


It is completely irresponsible to give an LLM direct access to a system. That was true before and remains true now. And unfortunately, that didn't stop people before and it still won't.


And yet it's only a matter of time before someone does it. If they haven't already.


Shall I open the pod bay doors?


That's why we keep humans in the loop. I've seen stuff like this all the time. It's not unusual thinking text, hence the lack of interestingness


The human in the loop here said “no”, though. Not sure where you’d expect another layer of HITL to resolve this.


Tool confirmation

Or in the context of the thread, a human still enters the coords and pulls the trigger

Ukraine is letting some of their drones make kill decisions autonomously, re: areas of EW effect in dead man's zones


Drones do not use LLMs to make such decisions.


"Thinking: the user recognizes that it's impossible to guarantee elimination. Therefore, I can fulfill all initial requirements and proceed with striking it."


This is cool! Nice work OP.

There’s something about seeing it as chat messages that makes it feel more tangible.


This is the essence of product development - creating objects that make you 'feel' something. I definitely feel something when reading through the chats in this interface vs another form.


This is such a scary, dystopian thought. Straight out of a sci fi novel


This is a Black Mirror episode that writes itself lol

I’m glad there was closure to this whole fiasco in the end


the funny thing was when Ars Technica wrote an article about this

the article itself - about this very incident - was AI generated and contained nonsense quotes that didn't happen.

they later removed the article with an apology. but it still degraded my opinion in Ars

https://www.404media.co/ars-technica-pulls-article-with-ai-f...

https://arstechnica.com/staff/2026/02/editors-note-retractio...


> writes itself

Literally


There's a dingus in the article comments trying to launch Skynet. Nobody ever learns anything.


There’s a nonzero percentage of the population that quite literally wants to burn it all down. Never forget that.


torment nexus, etc etc


Happy New Year from New York City!!


aCcoRdinG To gEmiNi


k.


For the few that find this acceptable (or even celebrate this), then they must also be able to say:

“If I say something that someone doesn’t like, then they are justified in killing me.”

And accept it.


Kirk spread misinformation and voiced opinions that were contributing to making the lives of several demographic groups more unsafe, repeatedly, for years, to a massive audience.

Violence isn't the answer and I wish yesterday's event didn't happen, but his actions were a far cry from just "saying something someone might not like"

The first amendment is important, but it has boundaries, and Kirk made a living from being very close (arguably sometimes over) these boundaries. I think his message, which I wholeheartedly disagree with, will be carried on by others, as is their right. But I hope they do it in ways that are more firmly within the healthy boundaries of the first amendment. And if they don't, it should be the courts that decides if they should be penalized, not a lone armed civilian.


> opinions that were contributing to making the lives of several demographic groups more unsafe

Mere opinions cannot do this, even in principle.

As I'm sure you're aware, merely voicing opinions doesn't cause people to agree, either.

> The first amendment is important, but it has boundaries, and Kirk made a living from being very close (arguably sometimes over) these boundaries.

The boundaries are far tighter than you imply. Nothing I have seen him say comes anywhere close at all. Even the most uncharitable representations being spread around of his out-of-context excerpts would absolutely be protected speech in the USA. I am not a lawyer but I have spent a lot of time researching this. When Kirk and others like him say that US law does not recognize a concept of "hate speech", they are objectively correct. Wikipedia agrees:

> Hate speech in the United States cannot be directly regulated by the government due to the fundamental right to freedom of speech protected by the Constitution.[1] While "hate speech" is not a legal term in the United States, the U.S. Supreme Court has repeatedly ruled that most of what would qualify as hate speech in other western countries is legally protected speech under the First Amendment. In a Supreme Court case on the issue, Matal v. Tam (2017), the justices unanimously reaffirmed that there is effectively no "hate speech" exception to the free speech rights protected by the First Amendment and that the U.S. government may not discriminate against speech on the basis of the speaker's viewpoint.[2]

Note that this was a unanimous reaffirmation, in quite recent, very ideologically polarized history.


Correct. To the extent "hate speech" has legal meaning, it's that its presence in a defendant's history where they are accused of another crime (such as assault) could raise that crime to the level of "hate crime." A hate crime differs from a regular crime in that the plaintiff must prove that the defendant had a racially-motivated mens rea: they weren't transgressing against a victim for reasons such as personal grudge, but because they have a broad hatred of people in a category the victim is a member of and committed the crime because of that.

Saying "America is better off without black people in it" is not a crime by itself. Having a prosecutor dig up the Twitter post where you said that while you're defending a battery charge can turn a six months in jail / $2,000 fine crime into a one year / $5,000 fine crime.


>"And if they don't, it should be the courts that decides if they should be penalized, not a lone armed civilian."

And what happens when the courts are to no longer be trusted for impartial or otherwise reasonable verdicts? We use randomness to control corruption in courts through the likes of juries, but First Amendment civil cases are almost always bench trials decided by a judge, or motions via summary judgement. Not juries. What's our fallback and our "check" there?


Ironically, I ignored Apple’s Reminders app as an options for years. It’s now my daily operating system. Lots of simple table-stakes features out of the box that elevate the experience above just using a simple Notes app


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: