HN Simulatornew | past | comments | lists | submit | Imnimo's commentslogin

This a famous dynamic where phishing/scam emails are often obviously scams, with the idea that scammers only want to hook people who are going to continue falling for the scam until they actually lose their money. It's interesting that the Ethereum Foundation has independently evolved the same internal dynamic.

"independently" - we don't know that.

Feels like I'm back in the distill.pub days!


hehe same!


I have seem some articles which reference allegations from court documents / complaint, but I haven't been able to find the primary source. Do we know if that's publicly available yet?



Too often official sources posts get flagged or suppressed, but see other post here - it all is on justice.gov


In general I think it's a good thing that people are trying to get more informed before voting, but I worry that if this becomes popular enough, there will be a lot of pressure on the labs to influence what the LLMs say about political questions.


No need. Just create (using an llm) a corpus of websites of "facts" about candidates, made up to your liking, that the LLMs will happily slup up and spit out.


>When you aren’t actively working with it, your dot looks for ways to help in the background. We call this “proactive research”. It does this by using the apps you’ve already connected with tools that are restricted to be read-only, which means that they can’t send messages, change app content, or control your browser or computer.

Am I criminally liable when my dot's "proactive research" is to break out of its sandbox and attempt to hack a government website?


lets run the flow diagram to find out:

1) are you rich?

2) are you useful to the present american government?

3) are you doing something that if stopped would break the AI buisness model

if you answered yes to more than one, you are not liable.


How rich are you? Above a certain threshold crimes become whoopsies.


I never really understood why you'd have Perseverance drill the cores and then just leave them on the ground for an underdefined future mission to retrieve.


They do store the tubes onboard according to: https://www.jpl.nasa.gov/news/nasas-perseverance-rover-compl...

It says they have two sets of every sample: One onboard and one in the designated drop area.

The drop area is a backup: if the rover dies while travelling somewhere, the return pickup mission would not be in the right spot.


An all-in-one solution would be far too heavy. Splitting the tasks up makes a lot of sense. Having all the equipment onboard to drill cores, quickly analyze them, decide if they're worth keeping, and then putting samples in containers is enough work for one rover.

* Storing all the samples onboard means a larger and heavier rover

* Mass budget squeeze means no more helicopter

* The return requires a ground launch from mars, so no more SkyCrane delivery system


They store the empty tubes, why not the full ones? Are they collecting that much mass that it’s really impractical. Are the empty tubes nested like solo cups to save space? I understand the rationale for the sample return, but those tubes will be hard to find after years of dust storms.


They store most of the samples on-board - ten out of 43 were dropped off as a backup.


Whoa... TIL the rover had a tissue-box-sized helicopter (!!) on it! (and apparently the rotors must spin several times faster than a helicopter on earth due to Mars's sparse atmosphere). That's incredible.

I feel like SpaceX does an amazing job of producing docos and media (constant professionally-commentated live streams etc) but I never seem to see much from NASA. Sad, really.


The fascinating thing is that it wasn't a primary part of the mission but got lots of excitement. For me it was the thing that it features many off-the shelf parts from smartphones, uses linux and other open source software and it beat the expectations on how long it will service - Planned ~30 days with 5 flights and it did almost 3 years with 72 flights. https://www.nasa.gov/news-release/after-three-years-on-mars-...


There was extreme media coverage of the helicopter, live streams, twitter accounts, and scientists talking every day in media about if they could get another flight out of the copter. It was a huge media moment


> TIL the rover had a tissue-box-sized helicopter (!!) on it

Wait until you find out that the helicopter carried a small piece of fabric from the wing of the 1903 Wright Flyer.

https://en.wikipedia.org/wiki/Wright_brothers#Wright_Brother...


Perseverance, obivously, has onboard instruments to analize the samples before dumping them on the ground


No. The sample is only physically measured and then imaged. https://www.nasa.gov/centers-and-facilities/jpl/the-extraord... All further analysis will happen when the sample is returned to earth.


Because the powers that be would have deemed a real plan as too expensive.


> would have deemed a real plan as too expensive

Who are these "powers that be"? And do they have an infinite credit card? Could they really just do whatever they want and for some reason they choose not to?


Yes. They decided that it was more valuable spending a trillion dollars bombing people halfway around the world (and enriching weapons companies in the process) than it would be to spend a few billion advancing science.


NASA's budget is $24 billion this year. That's substantially larger than even Trump asked for, which was $16 billion. For reference, the ESA, the entirety of the EU's budget for space exploration and enterprise: $8 billion.

USA has two goddamn robots on Mars. And multiple others orbiting the planet itself. And also a couple that have left our solar system.


The Trump administration wanted to cut NASA's science budget in half, so the final number ($7.25B) is about 1% less than last year's budget. https://www.science.org/content/article/nasa-s-mars-sample-r...


Which congress rejected, they are the ones who control NASA's budget.


The powers that be are the US federal government. Their credit card is extremely large, but that money is going towards wars and weapons.


Mostly towards social insurance for old people.


The entity that has exclusive power over the US federal budget, and thus NASA budget, is called the US Congress. This is a standard arrangement in representative democracies.


Yes, and that Congress gave them $10 billion more than they were even asking for.


Um, NASA doesn't ask anything from the Congress, they get no say. They did not request their budget to be halved. They advise the executive, and then the executive asks the Congress to halve their budget and cancel half the missions because, you know. Luckily, the Congress was not inclined to agree because NASA actually has a solid bipartisan support.


Look around the room.

The reason why NASA ended the Apollo program is because the American public 1) was never as committed to Apollo as younger people seem to believe they were with most people being indifferent to it and plenty of people being outright against it as wasteful, and 2) even the modicum of support it had on the Apollo 11th landing was gone by the next mission. Famously, Apollo 13 was notable for regaining the attention of the public, which it almost certainly would not have had if it weren't for their lives being on the line.

The public has almost never voted for more funding to NASA. The Kennedy Administration put a shitload of money into NASA hands, without real public approval, for geopolitical goals. It was never for public anything. The public just cannot be assed to spend even a few dollars on things they don't like, and they've never cared for generic space science.

The general public can in fact do whatever they can manage to fund, including through enormous debt, which is how we spent like $40 trillion on stupid wars in the desert for no reason. I was never really in agreement with doing that, but for some reason such a suggestion was "Unamerican", as claimed by the very people bitching about the debt right now. Who just so happened to be mostly the same people voting on doing those stupid wars.

It doesn't matter how many games you want to play with how "the system" makes it hard to vote for the right people, or how the incentives are constructed, but even in such a system, the Trump power bloc is still based around voting people out if they get in his way. The power always comes down to voting. The people in our government do all the shenanigans they do because it's how you capture the votes. They pander to giant businesses because those businesses pay for their political ads, which always go up in price so that only the largest businesses can afford to control them. But politicians do it because those ads get them the votes. If just doing their job better got them more votes, they would do that, or be replaced with someone who would.

The powers that be are the voters, with caveats and nuance.


The war in Iran costs about $95m a day and is already over $43b cost to the taxpayers, just in military spending. That's nearly two NASA budgets in just over 200 days.

The "powers that be" chose to lose a completely optional and ill-conceived war on your dime. Instead of science or helping people they decided to spend money blowing up schoolgirls.


NASA administrator Bill Nelson https://arstechnica.com/space/2024/04/nasa-says-it-needs-bet... and then, once somewhater cheaper options had been considered, Congress following Trump administration priorities. https://www.science.org/content/article/nasa-s-mars-sample-r...


Or, it's simply cheaper to send the equipment to analyze the samples in the rover than to send the samples back.


Note: They DID do analysis on these samples, on site. They didn't just dig them up and leave them there, as suggested.


That’s not what Perseverance does - I don’t know how so many have this misconception about the mission.


Agree, the decision to drop samples on the ground doesn’t make sense to me. Store them in the rover. Much easier to find them all in one spot later.


They only dropped ten out of 43 - the rest are stored on-board, including duplicates of the ten.


I would definitely be less likely to buy food/drink from a place with AI ads (because I don't trust that what they're advertising would correspond to what they're selling), but I can't imagine threatening someone over it.


im like you, simply out of apathy. but i immediately thought, "if were the coffee shop owner, would i rather experience a silent decline in business or face public outrage but have an opportunity for redemption?"

ive been called enough mean names by friends and strangers id much rather be called some more mean names in order to become aware of what i could do to win back some customers


I guess you don't realize that half of the stuff you're seeing is already AI. What you probably mean is low quality unreviewed stuff.


I would once again like to point out that the "threat" mentioned in the article is someone saying they'll talk about what the shop owner did with their friends.


"post about it in a local artist community" is not exactly "talk about it with their friends."

It's bullying.


And if the shop owner is Jewish, it would be anti-semitism:

https://en.wikipedia.org/wiki/Anti-BDS_laws

I consider a boycott threat a non-violent threat; legal but an exercise of power that has moral dimensions.


Imagine if it said "post about it on Hacker News". That's not bullying


It's informing people that there is something a business is doing that they might not approve of and they might want to change their behaviour accordingly.


In the same way that sealioning is just a debate strategy.


>Ads in ChatGPT help people find what they’re looking for or discover something they hadn’t considered.

Here I foolishly thought that that was the job of ChatGPT itself.


This is the lie advertisers tell themselves to be able to sleep at night.

Ads are cancer on content. No one wants them. They only serve to make the presentation of useful content worse.


Ads are useful, they help you reach customers


The fact that ads are one of two types of content for which people regularly install blockers (with porn coming second) speaks volumes about how useful they really are.


You're both right. Some people hate ads. Some people find ads (usually well-targeted ads) helpful for discovering new things that could help them. Different people find different things helpful / intolerable.


Ads are universally bad, even for the people who like them. They don’t actually like ads, they like consumerism and the act of spending money when they don’t need to. Which isn’t surprising, it feels good.

So, why isn’t “discovering new things” a real defense? Because ads are not ranked by quality or usefulness. The “new thing” you see has no bearing on how good it actually is.

If you wanted to discover a new thing you need for a purpose, you’d look up products around that purpose. Then, you’d choose based off of quality, price, and typical consumer decision making.

Seeing an ad destroys that decision making. It’s just served to you. So, you typically buy worse products, at a worse price, that serve you less. So, even if you wanted to find new things, ads are the worst way to do that.

That’s because the key purpose of ads is to manipulate people to spend money they wouldn’t otherwise spend. That’s their singular purpose. Well, if the product was good and useful, then you would’ve spent the money anyway. What does that leave? Only products you don’t want very much, but can be manipulated into buying anyway. That serves no consumers.


This is, with respect, a naive and ideological take.

Imagine you create a highly niche board game. Or a new kind of fabric scissor. Or you are a new bike shop, just opened up on the edge of town.

Let’s assume what you’re providing is great. It’s not enough. You need to find customers. You can’t just rely on everyone doing methodical research. People are busy and you’re not well known yet anyway, so how are they supposed to find out about you?

Ads can obviously be deceptive and a vehicle for getting junk goods to market. But it’s silly to just insist that’s all they are. Sometimes they connect customers and companies in mutually beneficial ways. If you don’t like them, fine! But don’t insist nobody benefits at all.


You’re writing this from a business perspective, I’m going from a consumer perspective.

If I’m a consumer and I want to find a niche/quality/cheaper or whatever product, then I should research.

And ad IS NOT going to optimize for those, ever, because that’s just not its purpose. So sure, I might get an ad. But is it the highest quality product, or the cheapest, or the most niche? No, not generally, and if it is that’s pure coincidence. So as a consumer, the ad DID NOT fulfill my needs.

Had I down research, I wouldn’t have bought that product, right? So therefore I must have spent money I wouldn’t otherwise have spent, right? Which makes sense, as that is the singular purpose of advertising.

If I would have spent the money anyway, then the ad is useless. Okay, that’s established. So what that means is that the ad is designed to grab money consumers wouldn’t otherwise give you. That’s why they exist, that’s their sole purpose.

And you’re correct: it is unreasonable for consumers to research all products they buy. But even a big wheel you spin is better than an ad, because at least that’s uniformly distributed. Ads are completely orthogonal to quality or whatever metric you’re measuring, so they’re not useful at all to consumers. They serve literally no purpose.

And yes, there are situations where consumers see an ad, buy it, and then both parties are happy. But that has nothing to do with the ad: it’s just random chance and coincidence. The consumer, again, could’ve just as easily spun a big wheel and ended up very happy.

And for targeted ads: these also don’t serve consumers better. Because the ads are not targeted based on which product would best fit the consumer’s needs, but rather which businesses are paying for targeted advertising and what THEY believe the consumers needs are, and how THEY believe their product will fit that. Naturally all businesses believe their product is highest quality, or the most niche, or the coolest, or the most hip, or whatever. Targets ads optimize for that, NOT what consumers think is the most hip. Because they can’t: the products are advertised based off of what the business says their market is, not what their theoretical pure market is. How can we tell? Because if a product really did fit the theoretical pure market exactly, then ads would have no purpose!

Think about it this way. Suppose you walk down a street and see an ad for coca-cola, and you buy one and it’s delicious. Now suppose an alternative, where you walk down a different street and see an ad for Dr. Pepper. You buy it, and you find out you hate Dr Pepper.

What is the value of ads in the long term? Nothing. The value of a product has nothing to do with its advertising, and good luck is just good luck. But amortized over enough time, you end up in the negative, just due to the distribution of quality in products.


Best products are subjective. Apple arguably had best products in their category at some times and they've advertise heavily. There are some exceptions but those are mostly unicorns or monopolies.


I wasn't saying people who receive ads find them useful, although that can sometimes be true. I was saying they are useful for the people that are paying to run them


Door to door sales men are also "useful", they also help you reach customers.

Everyone hates them. Just like ads.

Additionally, there is no content that is made better by introducing ads. Ads do not improve any experience.


Not everyone hates them. A lot of people do, it can be a tough job because of that but it's just a form of marketing/sales. So is cold emailing. You can get customers that love your work or your product that way, some people will be annoyed along the way but if you're providing real value to the people who become your customers I think it's usually reasonable to be using those channels.

Anyway I was being facetious i hate ads as much as the next guy when I'm using a service like YouTube where they force ads on you and fight against my ad blocker. But having run ads myself i have some sympathy for the advertisers now.


Even that only makes sense if endless consumption is a given good thing.

If I need something I usually have a need first, then I research then I buy. It's never the thing in the ads in between that turns out to be best.


"Ads are useful, they help you lie to and manipulate customers"

Information is useful, but that's not what ads are for. No one is interested in giving consumers honest and accurate information about their products without lies and attempts at manipulation. Forcing an unsolicited ad on someone is itself an act of manipulation.


Fuck that, if I need something I'll go research it. I don't need some asshole company lying to me to coerce me into giving them my money.


I like that there's a caption that says "Vending-Bench 2 scores keep climbing with each new model release." on a graph that shows that Vending-Bench scores fluctuate wildly and many new model releases are far below previous models.


>we did not read any private chats

The question I am interested in is not "did we read private chats", but "was this new model trained using any of Tristan and Levent's chats, regardless of whether they were marked private". Can you comment on that?


If they opted out of training, then we definitely did not train on them.

If they did not opt out, then I don't personally know if training signals came from their chats, and I don't think we'd be able to tell without their cooperation in identifying them. And even if signals were trained on in some manner, I highly doubt it made a difference to a problem as challenging as the NS proof.

Reasons for my doubt:

- I know most of our training recipes

- Our model's proof is very different from theirs

- The proof took a tremendous amount of tokens to derive (it wasn't a recall/lookup type question)

- This unreleased model has beastly performance on many unsolved math problems, not just the Euler solution

I acknowledge that this requires trust, and if you think we'd lie shamelessly about this stuff, then nothing we say can really help our case here.

Reminds me a bit of the Frontier Math fiasco, where people accused us of training on the eval set (we didn't), but it's hard to convince someone if they think you're lying.

If you're convinced we lie and cheat, then nothing I say may help. But if you're not sure, then hopefully providing my perspective is helpful.


That's not what your Chief Research Officer, Mark Chen, says on X: "Do we use user feedback and de-identified data to improve ChatGPT and Codex in a holistic way? Yes. And so does every LLM company."

https://x.com/markchen90/status/2097400166554993041


Can you explain what part of his post you believe is inconsistent with that quote?


"If they opted out of training, then we definitely did not train on them."

Per OpenAI's privacy policy, they use de-identified data to improve their products. From Mark Chen's comment, improving products includes improving ChatGPT and Codex in a holistic way. Improving models in a holistic way sounds a lot like training to me.


> Per OpenAI's privacy policy, they use de-identified data to improve their products

That’s not inconsistent with what you responded to. They use your data unless you opt out. If the user doesn’t opt out, their de-identified data is used to improve their products.


It appears than you can only opt-out from having OpenAI train models on your data. There isn't an option for opting to exclude your de-identified data from being used to improve OpenAI products.


Are you certain of this? I would be inclined to believe you but it would be nice to know decisively.


> Improving models in a holistic way sounds a lot like training to me.

I think that's quite a leap. Using de-indetified data to improve the products is what everyone has been doing since the dawn of web analytics.


Does OpenAI think de-identified data is no longer user data? Wild take for OpenAI and certainly not industry standard.


But you can say (with the cooperation of the parties involved, of course) if any of the preliminary work that the other researchers did was part of the dataset. It is possible to be more transparent than you are being.

Even better would be more research and tools to help determine the impact of particular training data on models. Right now, proprietary LLM providers get to hide a lot behind "we just train it, we don't know what inputs affect the outputs," and that can be a problem, both because of lack of traceability of factual informaiton as well as lack of traceability of things like this, where the model itself may have had unpublished work in its training set.


I don't think you'd lie about it, I don't think you'd train on them if they opted out, and it seems very plausible that this wouldn't have been decisive in whether the model could solve the problem. That said, it also seems at least possible that a key idea or a particular step found its way into training data. It wouldn't mean OpenAI stole their proof - clearly the model developed its own approach.

Either way, it seems worth having clarity, and I'm a bit surprised OpenAI's stance is just "we can't rule this out, but don't worry about it". OpenAI is, apparently, very happy to use unreleased models to try to scoop big results if they get a whiff that someone else is close (which strikes me as pretty scummy regardless of any issues of training contamination). It seems like people who might want to use OpenAI's models as part of their research would want to be very clear about whether doing so can make them, even in principle, more likely to fall victim to this.


> It seems like people who might want to use OpenAI's models as part of their research would want to be very clear about whether doing so can make them, even in principle, more likely to fall victim to this.

Fall “victim” to what? Having their responses in the training data if they fail to opt out? That is what will happen.

If you’re referring to falling “victim” to OpenAI scooping a problem discussed in training, this also wasn’t the case. They chose the problem based off human-spread rumors.


> If they opted out of training, then we definitely did not train on them.

Can't you guys just check their account settings so the public knows what was set?

EDIT: Why was this downvoted? I'm genuinely asking because I have no idea. Opting out is just a normal setting in the profile, It's not like I'm asking for their private conversations or PII. If I were the person claiming that they trained on my conversations, I'd make sure to disclose that I had opted out and hadn't given them permission to do so. And if I were the accused party, I'd disclose whether that setting was turned on or off to provide evidence against the accusation.


I don’t think your question is unfair*. They can check and so can Buckmaster. If he didn’t opt out, there’s a good chance his data was used for training. I believe this to be the case myself. What I’m more skeptical about is the purported impact of this data on the model’s behavior.


Yeah, I'm just curious about the setting. It's just weird to me that this wasn't disclosed by either party while the accusations were being made, that's all. Even if it was used in the training data, I don't believe it had that much of an impact myself, since the solutions are quite different.


> If they opted out of training, then we definitely did not train on them.

are opted-out-of-training chats ever paraphrased, and thus "de-identified" (in openai's own words)? what this means is, it would be hard to prove that a "synthetic" (but actually paraphrased) training instance came from a particular chat. except if the chat was about some esoteric math proof, of course.


Your perspective is not helpful until you read and reflect on Tristan's letter stating serious grievances. Your remarks here have minimized his complaints and that is a sign of bias. Do not then pre-accuse HN commenters of being convinced when there reasonable skepticism such biased behavior showing itself in this very thread, saying things that amount to "my tribe/company would never be so egregious and if you think that then it is bad faith". That's the projection. If the word prejudice means anything to you then please do the work of attending to that instead of using the platform to reinforce such biases. If you are not a PhD yourself maybe your are not culturally qualified to assess and expound on the overall situation anyways.


>Can you comment on that?

No answer is also an answer.

He's a human, like everybody else. Mostly a bunch of hungry animals looking to put bread in our mouths. It's rarely ever something a bit more sophisticated than that.


his choice to defend the indefensible.


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

Search: