HN user
You both seem to be using a different definition of "singularity" from the one I'm familiar with. I've always understood it to mean a rapid feedback loop in which AI creates successive, increasingly capable generations of AI outside human control, rather than simply a level of technological advancement that would be incomprehensible to someone
And this thread is seemingly full of people claiming AI can read it while simultaneously sharing that AI could not read the actual message, only the decoy as demonstrated in TFA.
That’s 100% on the authors for failing to make the default main “hidden” text and the decoy easily distinguishable. The way this is set up is incredibly confusing.
I would be very willing to pay more! The choice between “you may get a correct answer, or you may get lied to, without a clear way to distinguish between the two” and “you may get a correct answer, or a clear indication that the answer was not found” is pretty clear. One is a much more useful tool than the other. I don’t see any real incentives for companies making LLMs to keep their AI factually unreliable. (Full disclosure: I work for one, but I’m definitely not in the rooms where such decisions would be made.)
The article is about how people decide to share links, nie whether they share links at all.
Maybe that’s because I work with agentic AI in my day job, but this seems utterly obvious to me: no reasonable person would ever claim that LLMs are better at keeping secrets or enforcing rules than human employees.
This notice is not about comparing humans and LLMs. It seems that the system was designed in the only reasonable way: with a deterministic permissions layer separate from the agent. But that layer failed to work properly.
So the notice is comparing the difference between how the system was supposed to work and how it actually worked in reality. Normal post-mortem stuff.
Do you mean things like system prompts or things in Claude.md?
All of it - system prompts, user prompts, few-shot examples, Claude.md, things that an agent learned by exploring its environment...
So when I /compact my session what's that even the equivalent of I wonder.
Sleep? :)
I like to dunk on Meta as much as the next guy, but I think this makes sense: deterministic verification like this is not, and should never be, the LLM’s job. The tools it has access to should enforce the permissions layer, ensuring that the LLM can never perform actions the user themselves should not be allowed to perform. In this case, the tool failed to do that.
Is it? Both supervised learning and reinforcement learning are ways of training the model, and the difference between them is not that big. I would say that innate means "in the weights", while non-innate means things the model learned during inference, during its "lifetime".
I think this is exactly it, but let me ask another question (which is not rhetorical, I really don't know). Does the fact that one can describe what consciousness is and where it came from in humans help them to detect it in non-human and/or non-biological entities?
And also "instilled during their reinforcement training", and we are currently pushing planning hard there, for autonomous agents.
This was said in the context of a person predicting a stock market bust, so of course a stock market index price is the relevant number here.
It’s obviously not a new model capability. But using this well-known, existing capability to solve this particular issue is only obvious after the fact.
It’s a useful trick to have in one’s toolbox, and I’m grateful to the author for sharing it.
Performing 40 songs in exchange for a property does seem like serious effort...
The top comment categorized scraping as abuse ("abuse such as [...] scraping") - that's precisely why some accuse its author of lack of self awareness.
Perhaps LLMs have mimicked the style because authors have popularized it and clearly it serves some benefit to readers.
Shouldn't the code say:
position = (position + direction + 1) % 12;
Just had the sitename put into the value of the cookie since, and never really needed to think about that.
How would that help? This doesn't seem like a solution to the CSRF problem
As a childless OMSCS graduate, I also can’t imagine doing it while having kids, because it took basically all of my free time. That said, I met quite a few people in the program who were in situations similar to yours. I have no idea how they managed it, but they somehow did.
In that case the winning strategy would be to switch hedge funds every 3 years.
When you flip a coin, you can easily get all heads for the first 2-4 flips, but over time it will average out to about 50% heads. It doesn’t follow from this that the winning strategy is to change the coin every 3 flips.
Okay? I specifically responded to your comment that the parent comment implied "if you make a mistake and say sorry you are also a psychopath", which clearly wasn’t the case. I don’t get what your response has to do with that.
I think the point of comparison (whether I agree with it or not) is someone (or something) that is unable to feel remorse saying “I’m sorry” because they recognize that’s what you’re supposed to do in that situation, regardless of their internal feelings. That doesn’t mean everyone who says “sorry” is a psychopath.
But this sounds like an ideal setup, doesn't it? Tim is fantastic at execution, but he does need a shot of big-picture vision every now and then. Tim as CEO with Steve as Chairman, steering the broader direction, feels like it could have been a perfect pairing. The issue with how things actually turned out is that Tim ended up on his own - all execution, no vision.
Good call. I would feel extremely weird seeing "[my full name] is still alive" as a title somewhere...
I think you misunderstood what people are taking issue with. You explain that this matter is complicated and non-trivial - and yes, that’s exactly the point!
People don’t have a problem with real-time communication via audio or video in general. They have a problem with the suggestion that it’s a trivial issue that can be easily fixed by "jumping on a quick call."
The point about there being a "fairly in-depth" description of the issues isn’t that there’s nothing more to discuss - fixing those issues would obviously require talking through the specifics. The point is that this is a real problem that requires action and commitment, so suggesting it’s a non-issue that can be clarified with “a quick call” comes off as dismissive and unproductive, whether that’s intentional or not.
There is a big leap between them not being the sole person responsible for technical decisions and them not even necessarily having a seat at the table for technology direction. The former is understandable. Later - quite surprising.
That’s like saying Volodymyr Zelenskyy supports Trump. Foreign politicians operate outside of U.S. domestic politics - they don’t get to choose other countries’ leaders. Their job is to use diplomacy to navigate international politics in whatever shape those politics happen to be in.
20 million unique users is not that much. I don't understand the claim that this constitutes 9% of all IP addresses. It doesn't. There are about 4 billion public IPv4 address. 9% of that would be closer to 300 million.
I don’t see a claim that anyone with a negative attitude toward AI shouldn’t be listened to because it automatically means that they formed their opinion on older models. The claim was simply that there’s a large cohort of people who undervalue the capabilities of language models because they formed their views while evaluating earlier versions.
Turns out what constitutes "claiming" an IP on the site is nothing like you’d expect. You don’t need to prove you control the IP. All it takes is embedding a transparent 1x1 tracking pixel on a website, and every IP that loads the page gets counted as “claimed” by you. In other words, it’s just a tally of visitors (or even ad impressions), not actual control of the IPs. So there’s really nothing meaningful here.
Why train the model to know how to use very specific tools which can change and are very specific only to ChatGPT (the website)? The model itself is used in many other, vastly different contexts.