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

If you don't review what the product does, you are irresponsible for the product.

Is the CEO responsible for a company's financial performance? Do they review every line of code the company writes?

It is more irresponsible to spend the time reviewing all of the code rather than spending that time on things with bigger levers for satisfying your customers.


yes but if a dev pushes a line of code that wipes the accounts of millions of users at a fintech, the dev will get fired but the CEO will get sued into oblivion. if the agent isn't responsible, you HAVE to be, cause angry people wont listen to "it's no ones fault your money is gone"

Put an ORM in your private repo which randomly 1% of the time calls DROP TABLE.

When I read SoaNM, I quickly realized that it had been used as a style guide for several later books, and most long-form tech articles in Wired, the NYT, etc.

The two errors, then, were that the LLM hallucinated something, and that a human trusted the LLM without reasoning about its answer. The fix for this common pattern is to reason about LLM outputs before making use of them.


A big problem now both internally to a company and externally is that official support channels are being replaced by chatbots, and you really have no option but to trust their output because a human expert is no longer available.

If I post a question to the internal payment team's forum about a critical processing issue and some "payments bot" replies to me, should I be at fault for trusting the answer?


I know this is happening with external customer support, but is this really happening internally at big companies? Preventing you from talking to a human in the correct department about an issue feels like a bomb waiting to explode.


There is at least an effect that chatbots have become the primary line and support, and even if you are not prevented from talking to a human, the managers of the humans you would talk to have decided that since the chatbot is there, it is inappropriate for them to be spending much time supporting coworkers in other departments when the chatbot can do it.

So to a degree, corporate politics can sort of discourage it.


I'm sure it is. Thankfully I don't work for a company this large any more, but when I was employed by a multinational with 30K+ employees, our IT department was outsourced to India and you had to get through a couple layers of phone tree/webchat hell to actually talk to a real person. I could easily see companies of this size replacing their support with LLM nonsense.


Teams are heavily incentivized to incorporate AI in their internal workflows. At Meta it is a requirement, and will come up in your performance review if you fail to do so.


Yes, of course, and the company which removes human experts should expect things to fail in the manner that things usually fail when you remove your internal experts.


> The fix for this common pattern is to reason about LLM outputs before making use of them.

That is politics. Not engineering.

Assigning a human to "check the output every time" and blaming them for the faults in the output is just assigning a scapegoat.

If you have to check the AI output every single time, the AI is pointless. You can just check immediately.


The humans are not scapegoats, because they are capable of taking on responsibility.

There is a point to using LLMs. They can save time by doing a first pass. But when they do the last pass, disasters will follow.


Well, I'd say there's two dimensions:

1. Check frequency (between every single time and spot checks).

2. Check thoroughness (between antagonistic in-depth vs high level).

I'd agree that, if you're towards the end of both dimensions, the system is not generating any value.

A lot of folks are taking calculated (or I guess in some cases, reckless) risks right now, by moving one or both of those dimensions. I'd argue that in many situations, the risk is small and worth it. In many others, not so much.

We'll see how it goes, I suppose.


Groooooooaaaaaaaaaaaannnnnnnnnnn


Well, attempts to engineer the brittleness out of human behavior have not worked, like, ever.


If "the level of awareness that created a problem, cannot be used to fix the problem", then you're asking too much if you expect a human to reason about an LLM output when they are the ones that asked an LLM to do the thinking for them to begin with.


This feels like a rediscovering/rewording of Kernighan's Law:

"Debugging is twice as hard as writing the code in the first place. Therefore, if you write the code as cleverly as possible, you are, by definition, not smart enough to debug it." ~ Brian Kernighan


It's an old saying, I think Einstein is cited most often for it... something like this according to Google:

"We cannot solve our problems with the same thinking we used when we created them."


In this case you would replace the human.


Yes, I'd fire them, and then hire a more competent human.

I'm pretty happy with the team I've built. They make solid decisions that I can trust every time. I can't say the same for the LLM.


However - Automation bias is a common problem (predating AI), the 'human-in-the-loop' ends up implicitly trusting the automated system.


At least pre-LLM automation was written by a careful human who's job was on the line, and was deterministic.


When organizational incentives penalize NOT using AI and firing the bottom x% regularly then are you really surprised LLM outputs aren't being scrutinized?


Yes, because trusting LLM output is a great way to be in the bottom x%.


It's more like, the LLM "hallucinated" (I hate that term) and automatically posted the information to the forum. It sounds like the human didn't get a chance to reason about it. At least not the original human that asked the LLM for an answer


I’m not in AI, but what is happening is that it is building output from the long tail of its training data? Instead of branching down the more common probability paths, something in this interaction had it travel into the data wilderness?

So I asked AI to give it a good name, and it said “statistical wandering” or “logical improv”.


If you don't like hallucinate, try bullshit. [NB: bullshit is a technical term; see https://en.wikipedia.org/wiki/On_Bullshit]

https://www.psypost.org/scholars-ai-isnt-hallucinating-its-b...


That is my preferred term, but it seems to derail discussions that might have otherwise been productive (might...the hope I have)


Placing a bet based on insider national security information should be regarded as leaking. But the markets aren't the problem here.


Well markets give a huge financial incentive for it. Before you had to get paid a bribe for an intelligence agent. Now you can just "legitimately" bet on a market. It's a LOT easier and more spread out, I imagine.


The existence of banks gives a huge financial incentive to rob them. That doesn't mean we should get rid of banks. It means we should create a huge disincentive to rob them (which we do). Same thing needs to happen to people using national intelligence secrets in prediction markets.


That requires active participation in regulation and enforcement.

If anything, this administration is moving the other direction when it comes to betting markets, crypto, investments, etc.

What you are saying is logical for how enforcement should happen, but it isn’t happening that way.


Banks are about saving money and reducing risk. Arguably when they got into high risk investments to make high reward is what contributed to the 2008 recession.

Not sure if banks are the best example of proving the country is not headed towards high risk gambling.


On the other hand, it provides the whole world with the information and not just some spy agency. Isn't that more fair system?

And people dying in question is army, professional murderers for hire themselves, not a big loss.


You couldn’t design a better system for incentivizing leaks if you were trying. Hell, the CEO literally said as much. Not sure how you can conclude the markets aren’t the problem.


Yeah I had to reread that part... I was like, no way the CEO of Polymarket publicly said on record that it incentivizes leaks. Had to check to make sure I wasn't on the onion.


The CFTC has been defunded and dismantled. The industries it regulated don’t even bother to put on a mask anymore.


He was talking about companies at that part of the interview, and getting company info leaked to the market is generally a good thing.


Wow, it took some time for me to dig the interview out(0). I think it's stupid that Atlantic did not link to it, and that they misrepresented the context.

I agree that company info being leaked is whatever. No one is hurt by knowing that Apple is working on a foldable phone; maybe an exec loses his million dollar bonus and can't upgrade his yacht this year, and the market can operate off of that knowledge.

But the flip side is that there's no way to distinguish between leaked company info and leaked government info, and up until this era of history, there was rarely financial incentive for anyone to leak govt info, and even if there were, it was almost impossible to do so completely anonymously.

I'm not necessarily agreeing with the article. Who knows if that actually happened? But the incentives make it more plausible than ever.

(0) https://www.youtube.com/live/t647EWQst5A?si=TPNx7RbxWvn9wVdn


Let's make bets on "xxxx leaked yyyy info", so people will be incentived to discover/denounce who cheated.


How do they know? Has this been published in a Reliable Source?


This is the official Wikimedia Foundation status page for the whole of Wikipedia, so it's a reliable primary source.


Actually, usage of primary sources is kinda complicated [0], generally Wikipedia prefers secondary and tertiary sources.

[0] https://en.wikipedia.org/wiki/Wikipedia:No_original_research...


Yeah, but the purpose of an encyclopedia like Wikipedia (a tertiary source) is to relatively neutrally summarize the consensus of those who spend the time and effort to analyze and interpret the primary sources (and thus produce secondary sources), or if necessary to cite other tertiary summaries of those.

In a discussion forum like HN, pointing to primary sources is the most reliable input to the other readers' research on/synthesis of their own secondary interpretation of what may be going on. Pointing to other secondary interpretations/analyses is also useful, but not without including the primary source so that others can - with apologies to the phrase currently misused by the US right wing - truly do their own research.


If you spend any time on Wikipedia, you'll find that secondary sources from an existing list are always preferred. The mandate from the link in GP (https://en.wikipedia.org/wiki/Wikipedia:No_original_research) extends, or at least is interpreted to mean to extend to, actively punishing editors who attempt to analyze or interpret primary sources.

My original post was a joke about this.


https://www.brookings.edu/articles/macroeconomic-implication...

"There was a significant drop-off in entries to the United States in 2025 relative to 2024 and an increase in enforcement activity leading to removals and voluntary departures. We estimate that net migration was between –10,000 and –295,000 in 2025, the first time in at least half a century it has been negative."


Look at the number of covid deaths in countries other than the US and consider updating your news diet.


You do realize that the US had _one of the highest_ per capita Covid deaths amongst developed nations?


The correlation between mortality and body mass index is striking.


Maybe the President should have taken that into account when lying publicly about the impacts that he admitted in private conversation, or mocking and undermining expert advice?


[flagged]


Fauci injected money into the economy? Who was the President at the time?


US has one of the unhealthiest populations amongst developed countries too, so maybe it’s not that surprising.


Excess death from Covid is a non-trivial topic. Sweden had a very different approach to covid response, and yet had a very average number of excess death. The post-covid investigation provider some clear insight of what was primary causes to excess deaths, and yet very little of those conclusions has became common knowledge.

The primary group that had excess death caused from covid was to people living in homes for elderly care, and the primary cause was a lack of initial process and gear by people who worked at those locations. They were not given enough time to keep up a higher standard of sanitation (often given less than 15 minutes between patients), and protective gear was lacking. They also heavily depended on mass transportation which was a primary location for the virus to spread. A better early response in that sector, including shutdown/restriction of mass transportation would had saved many elderly people from early death.

To note, this had nothing to do with masks, vaccines, or shutdown of schools, which is the main points usually brought up in popular discourse. Sweden would have had one of the lowest number of deaths, with the exact same use of masks/vaccines/shutdowns as it did, as long as the response in elderly care had been done better.



The thing which distinguished him was getting good guests, before the hype hit. And he generally asks good questions and then shuts up while his guests talk.


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

Search: