"The commitment to the model, and subsequent inference proofs, hide the weights of the model; hiding the architecture of the model is not a goal of this system."
People will look at this and think "bad lawyers, you should not do this." But please note that this is literally how the advertising (or, e.g., entire social media) industry works: humans are subtly (or not so subtly) manipulated to do things advantageous to third parties.
Some might respond that "advertising is not illegal," but this argument is tantamount to saying that you should expect ads to be minimally faithful and maximally manipulative. In other words, economic/ social participants can reasonably do anything within the allowance of the law (or worse, wherever cost/benefit makes it optimal). Eventually the absolute disregard for truth or any sort of higher level value was going to give way to our present state of corruption. AI just makes it easier.
100% this. Marketing should be regulated so it's purely factual. No deception. Here's a product. It does this thing. If it does it better than others, show exactly how it differs but don't say it's better (because that's subjective). Don't try to trick me into believing it's something that it's not.
Marketing should also come with stronger commitment. If it says it will do something in a certain way but it doesn't quite, consumer protection laws should be a lot stronger than they are even in Europe.
What does “purely factual” mean? One might argue that having a colourful logo, and a picture of some bikini-clad ladies drinking a bottle of your product on the beach, does not constitute a factual claim.
That is easy. Were the intentional elements of the advertising intentionally truthful and meant to accurately represent the product?
You know how they currently spend lots of money designing and focus grouping to see if the average consumer will get the wrong idea about the product? They have to do the opposite. If they spend lots of money on the advertisement (enough that you would reasonably focus group, etc.), then you need to focus group until their impression of the product is something you would not disagree with in a court of law.
Why did they show people having fun drinking a soft drink on the beach? Was it to deceive the consumers into believing something false or having a false impression?
Why did they do that over a purely informational advertisement depicting the soft drink and explaining the taste and ingredients? I would consider such a advertisement clearly acceptable.
They have to demonstrate why their advertisement showing people having fun on the beach is intentionally truthful and non-misrepresenting relative to the consumer impression following the advertisement. I am sure they have focus groups who can answer your question. Ask them.
They have to demonstrate why their advertisement showing people having fun on the beach is intentionally truthful and non-misrepresenting relative to the consumer impression following the advertisement
Sounds like a court battle that could last for years and years. And in the end I think they would win for the simple reason that they can defend the advertisement by saying the image of people having fun at the beach and enjoying their drink is the intended use of the product.
And this is supposed to be easy? There are millions of companies advertising out there and the government is going to pursue these long court battles against all of them to prove their intent?
Proving intent to deceive is almost impossible in far more straightforward situations. What you have proposed is ridiculous, draconian, and would not be at all effective without essentially shutting down all media and installing a censorship bureau. So much for free speech!
Yes, this is very easy. “They can defend the advertisement” is not the standard I was proposing. That is the existing standard which, as you point out, is functionally impossible to fail. As long as you can create a tortured interpretation in the worst of all possible worlds that even a genie would be proud of, you get away with clearly intentionally deceptive statements.
As we know, false commercial speech is already, rightfully, not protected by freedom of speech. However, the standards for such have been so thoroughly blinded for the benefit of liars to the detriment of society as to fail to detect attempts to deceive so obvious that even children can tell let alone more subtle attempts.
Instead, the standard should be strengthened. Commercial speech should need to affirmatively “prove” it is non-deceptive and safe for consumption in reasonable proportion to the resources allocated to its production. Commercial speech should become more truthful and accurate and immune to misinterpretation the more resources that are allocated to it unlike the status quo where is becomes more deceptive.
This is a standard that is trivial to achieve for anybody with interest in the truth. Dry, informational statements that err on the side of caution are obviously acceptable. If you are producing a advertisement at substantial cost then instead of using your focus group budget on making sure the targeted consumer is thoroughly deceived, you will need to use it to make sure they only believe things you would also affirm in a court of law.
I colloquially call this the “Technically, your honor…” standard. If you say that you should probably lose.
This is not to say that the standard is that nobody in the entire world comes to a false belief, just that the overwhelming majority do not.
Reasonable defenses would be:
1. The false belief is not beneficial to the marketer.
2. The false belief is so small as to be immaterial.
3. The person with the false belief is not reasonably in the target demographic. (Nuanced statements might be misinterpreted if not in target demographic)
4. The interpretation leading to the false belief is unreasonable. (People can just be unreasonable and stupid)
5. The false belief is unexpected and could not have been reasonably detected given the resources allocated. (Minor mistakes and blindspots are acceptable, you can not account for everything, but you must account for more as more effort is used).
I'm not that interested in the fine details of your standard. If the burden of proof is on the advertiser to prove their commercial speech is acceptable, then the big question that follows is:
Who is the judge of their claims?
* Is it a judge in a courtroom? That will never work, as the courts are already overloaded and you're talking about adding hundreds of millions of additional cases to the docket per year.
* Is it a government regulator? That will make the Patent & Trademark Office look like a lemonade stand. You would need tens of thousands of people reviewing ads every single day.
And how do you presume to keep this regulator from being "thoroughly blinded for the benefit of liars to the detriment of society?" All regulatory bodies are subject to regulatory capture pressures.
The same one it always is, the finder of fact for that legal proceeding. This is really basic stuff. It is even more absurd that you did not even have the standard finder of fact amongst your options.
> That will never work, as the courts are already overloaded and you're talking about adding hundreds of millions of additional cases to the docket per year.
That is possibly the most inane argument I have heard in a long time. Adequate judgements against now illegal behavior will strongly disincentivize behavior relative to the past where it was legal. If running afoul of the new law resulted in a 100% of yearly revenue fine do you really think everybody would be chancing those odds resulting in hundreds of millions of additional cases? Obviously not; then my point is proven. Of course what you would actually do is scale the fine until it is commensurate with the desired degree of disincentivizing or until the undesirable behavior becomes unprofitable; the unreasonably gigantic fine I used as a thought experiment above is just there to demonstrate that your point is utterly inane.
The same one it always is, the finder of fact for that legal proceeding
You have already claimed above that the burden of proof is on the advertiser to prove that their advertisement is legal. This is a guilty-until-proven innocent system, so ALL advertisements will be part of legal proceedings.
Advertising is ubiquitous. You want to create a regime where all advertisements must be submitted for approval before publishing. You have not demonstrated how this could be made remotely practical, given the enormous volumes involved.
The nutrition information is highly regulated in both countries. The Japanese take it a step further and regulate the marketing part of the label. Pictures must match the contents in size, shape and appearance. Juices can only show photos of fruit if they are 100% fruit juice. Etc.
Not a novel topic on HN but marketing is by definition manipulative - regardless of the well-meaning of intentions. But IMO regulation is maximally worse since it incentivizes advertisers to maximize for legality for the competitive advantage rather than liberally interpret in favor of their product and customer.
People need to be aware of their own behavior in the light of attempted deception and choose "good" accordingly rather than being forced to accept marginally bad advertising merely because it's legal.
Advertisers will keep push the boundaries forcing laws to tighten continually - a perpetual cycle (as we can clearly see from books and books of laws, regulations and rulings). In contrast a leanly regulated society chooses to buy good products regardless of advertising will generate better results on all fronts. More laws is rarely the answer. Better people and morals are and laws can never generate that outcome. The founding father John Adams already wrote "Our Constitution was made only for a moral and religious people. It is wholly inadequate to the government of any other." - 1798
TL; DR be kind rather than forcing [fake] kindness.
Sounds so nice, for anyone who hasn't seen actual advertising in today's world, where only the absolute more clear and demonstrable lies are even theoretically illegal.
So we should build agents that read the ads, cut the crap, assess alternatives and recommend. If our brains are so easily hackable, then we should clean and filter our inputs.
The federal ministry of truth is called the court and already exists and prosecutes people for stating false things. How is slander enforced for example?
Courts don't decide truth, if they did the bar would not be "beyond a reasonable doubt" and we would never have wrongful convictions. They also wouldn't need juries.
Finding something to be true beyond a reasonable doubt is called finding truth for normal people. Otherwise, truth would only really apply to mathematical proofs, right? Is it true that the earth is roughly round? I don’t know for sure, I could be living in a simulation in an entirely different universe for all I know. But I would still say that the earth being a roughly round shape is true.
Truth wouldn't apply to mathematical proofs either, no. Truth is effectively undiscoverable and that's kind of the whole point, see the scientific method and consensus vs truth.
The problem with making the government decide what is true is that they have to codify the definition and how to identify it. If they can go with gut feel or what seems likely that will eventually be abused by those in charge.
The question was how false marketing laws would be enforced if there’s no ministry of truth. This isn’t about how the law is made but how you then decide what is true or not. And my point is that this is already happening for existing laws.
Courts don't decide truth and we shouldn't ask them too. Its bad enough that we already have laws that are vague enough that its only decided after the fact whether what you did broke them, one should be able to know what crosses the line so they can avoid it. Asking courts to be responsible for determining truth after the fact just makes that worse.
Having doubt in a piece of evidence is not saying all other pieces of evidence are true. Truth is effectively impossible to prove as you never can know what you don't know.
So let’s say A publicly says that B cheated on his wife and B sues for slander. A then states in front of the court that it’s true, which is an absolute defense. How do you think the court is going to decide that without deciding on the fact of the cheating?
Courts can't decide that B chested, how could they? The court can only rule on what evidence is provided, what laws are on the books and how to interpret them, and whether the evidence makes a strong enough case.
It seems you and I have a fundamentally different definition of truth.
So if the court decides, based on the evidence provided by both parties, that B did in fact cheat and A isn’t convicted, that’s not a finding of fact for you? Does that not explain how a court could handle the proposed advertisement laws as well? It’s pretty much the same procedure as is already used for existing laws like my example.
Another question so I get what you’re saying: If I build a radio that emits outside allowed frequency ranges and the FCC bans me from selling it, who is determining the truth wether the radio does in fact emit outside the allowed range?
You're swapping in the words fact and truth which are different. If a court years a case and makes a ruling following proper procedure, that is legal fact. That doesn't make it true.
If court decisions were always determinations of truth then wrongful convictions would be impossible.
Edit as I meant to answer your last question. The FCC almost certainly defines how they measure radio frequencies and is defining what measured frequencies are allowed and disallowed for transmission. I will say this question seems a bit like a straw man though, the realm of fact is much much larger. If a government must determine truth it often would not be based on things that can be isolated and tested.
So how does any of this matter for the proposed law on false advertisement? The court will still find the manufacturer to have distributed false advertisement, and will convict them. You might not call it truth but I don't see how that wouldn't work, especially since slander and similar laws already exist.
I remember being a young child when my parents were shopping for a car. I heard a radio ad for a car dealer claiming it was the best place to get a car, and I was excited to tell them.
I think that experience helped blunt the ability of ads to manipulate me ever since. I recognize, however that advertising has increased in sophistication, so I block ads pretty aggressively and try to avoid media that makes it impossible.
Advertising is one thing. Advertising in a court filing is a whole different deal. Lawyers are expected to advocate for their client as hard as they can, but they are expected to stay within the bounds of the truth while they do. ("Duty of candor" is written into the rules of most or all jurisdictions, at least in the US.)
If lawyers are going to straight-up lie, then that can cost them the case, or get them sanctions, up to being disbarred from practicing.
My point is not about what or where the ads occur. My point is that if you legitimize some ruthlessly selfish activities, then you cannot be upset when other ruthlessly selfish activities emerge.
> My point is that if you legitimize some ruthlessly selfish activities, then you cannot be upset when other ruthlessly selfish activities emerge
You can, in fact, be upset when lawyers abuse the system.
Not only can you, but you should.
These attempts to reduce all things across society into oversimplified, vaguely defined buckets like “ruthlessly selfish” and then saying we can’t be upset about unrelated things because they all reduce to the same vague bucket is the weirdest form of whataboutism. The argument doesn’t even make sense, it’s just a forced false equivalent to start complaining about a different topic (consumer advertising) in a thread about lawyers doing prompt injection in a court of law.
Is this really a plausible theory of how advertising works? Much advertising really is just informational (or the thing where you need to advertise your company under its own name on Google), but if the ‘manipulation’ stuff is so straightforward, couldn’t feeble-minded people such as myself ‘advertise’ to themselves to behave more in ways they otherwise wouldn’t? I’ve come round to the argument that a social element (of knowing that many others have seen the same ad) is important.
But those are professionals, expected to have a high amount of training, and exactly on the profession that decides if things are ok or not. They should know better, and those are bad lawyers.
I don’t mind advertising in news papers or on buses. I don’t like advertisements based on my profile created by collecting private data from me and stalking me through the whole internet. So advertisement is ok, collecting data and stalking is not. It’s not even working in a good way.
I do mind advertising on public transit. I'd rather not be subjected to looking whoever paid pennies to put up whatever they want. It's undignifying and usually doesn't even provide the agency very much money at all.
I grew up in a town with strict limits on commerce. No billboards, no sidewalk sandwich boards, business were required to offset from the road, plant shrubs / trees to reduce the view, both the building and road signs were required to be made of ‘natural’ materials and colors like wood and stucco with minimal branding color ‘pop.’
This was one of the most visually stunning and peaceful places I have ever seen in the world let alone these United States - and there is not a a single reason we can’t all enact that law at a national level. We just don’t.
Of course this doesn’t begin to touch on the stalking they companies get away with on the internet.
That's a funny way of saying "I can't think up any arguments against this position, so I'm just going to commit an ad hominem against a wildly diverse set of people for no reason other than to perpetuate flamewars and/or virtue-signal".
Can you please stop acting like a complete jerk on HN? This style of commenting, which is full of sneers, swipes, barbs and low-substance takes on various topics, is absolutely not what HN is for. When accounts continue to do this, we have to take note of their signaling that they have no interest in using the site as intended, and ban the account. This place only exists because enough others make the effort to raise the standards rather than drag them down. Please read the guidelines in full and make the effort to observe them if you want to participate here. https://news.ycombinator.com/newsguidelines.html
When I need a new car, I will look for one and compare available models based on their qualities and prices. I don't need to be passively bombarded by propaganda, daily and for years until then.
Advertising is not the reason I buy cars... The world was made car-dependent to maximize the profit of the automotive industry. Today, you need one to exist as a full citizen, that's a separate issue to advertising.
What? What I'm thinking is "bad judges, opening the gates for AIs to control us, and letting young people that could be interns have to make ends meet by less legal means. Those lawyers are heroes!"
Both accidents involved Iryo, a Italian operator that started running high speed trains on Spain a few years ago. Given how small their fleet is, it's weird that they have been involved in every recent accident, but at least the Adamuz one was clearly the fault of the tracks and not the trains.
Spain's network liberalization is recent-ish, operators like Iryo exist since 2021.
Perhaps the synchronization between renfe and others is a weak point in the system. Decades of assuming that everything in the network is theirs, increased wear and tear, etc.
To be fair, what you're describing is literally an increase in inefficiency caused by liberalization. It's what large corporations solve by "vertical integration" or "in-housing".
If you operate a rail network you need to keep track (pun not intended) of the exact state of all parts of the infrastructure, maintain and update the schedules and routes to account for repairs or disruptions, observe utilization and demand changes (not all of which can be profitable due to externalities like the social benefit of connecting people in remote places even if they don't provide enough revenue) and of course you need to keep track of where your trains currently are and where they are going and need a system of signaling, failsafes and communication to prevent incidents like "rear-ending" a stopped train (which this looks like it was).
Once you start breaking this up into a separate companies (like Spain has done and other countries did in order to "liberalize" the natural monopoly of public transit) you create inefficiencies and misaligned incentives. If you then throw competitors in the mix (who of course only compete in the few profitable aspects like high-speed rail because they only exist to extract surplus value from the system by withdrawing more from it than they put in - that's the literal definition of "profit") the misalignment of incentives gets worse and at worst so does communication.
As a software developer, it's kinda like having to build a publicly document general purpose API for your server-client application so it can work with arbitary third-party clients while having your team at the same time be moved into separate organizational units with different managers. Yes, sure, you can do that, but you better hope this will be reflected by another digit being added to the budget because not only will it make the project a lot more expensive but the same goes for every change that needs to be made to it in the future (not to mention front-to-back performance optimization is pretty much off the table) - except the point here is cutting government expenses so you don't get a real budget and this isn't a greenfield project but a brownfield legacy codebase never intended to serve this use case.
That should never cause a crash. You don't let one train go onto a section of track unless you know that there's no other train on that section of track. If train A has failed to move when instructed, then train B waits for it to move.
That all is true, and furthermore the AVE train was running way above the 30 kmh limit set when the automatic control systems fails. The only way this accident could happen is if someone violated every existing procedure.
The train operators do not seem to be a factor in either incident. Still too early to know what happened today, but the first findings point to an entirely different type of incident.
Also it's of course worth acknowledging that the means of preventing accidents as a train operator is fairly limited beyond the basics. Even if you see something and react in time, trains are extremely efficient at going where the rails will take them but that makes them also very bad at stopping to do so even when you want them to.
Google's goal is not to provide search; it is rather to gather infinite data about people and monetize it. If google offered a search API, it would destroy its surveillance system.
We should have outlawed targeted advertising and Google should have been broken up under antitrust a long time ago.
A lot of the mech interp techniques are glorified notions of probing. If models were generally interpretable with respect to the things we cared about, we'd know know when they're hallucinating, for example.
Identifying that a common neuron or set of neurons fires for golden gate bridge can be done only because you already know that you're looking for golden gate bridge. If models had recovered some new truth (e.g., the silver door road), you wouldn't be able to elicit it without knowing to look for it a priori.
Typically the analysis works in the reverse direction: we have a neuron (or SAE feature, or whatever) and we look at examples of data that makes it activate, then inferring an interpretation. This is a lot harder than probing for a known concept, and one of the points of this article is that if you try harder to understand the activation pattern, you will very often find that there is something interpretable there, even if at first glance it looked random.
This was one of my intuitive concerns reading this post.
I've done work with trying to train explainable models for e.g. making diagnoses based on neuroimaging, and it is not clear to me that if you applied these techniques to e.g. vision models trained exclusively on such brain images, that these visual representations or conceptual clusters and relations would really help understanding at all.
The main issue would be that you'd have no clear way to verify any generated explanation for such training data. I.e. we can "verify" some of the explainability shown in the article because clear and recognizable image clusters and/or semantic groupings pop out, because we already understand the domains. But if this were fMRI or e.g. EEG signals, we really wouldn't have that, so the explanations would themselves just be more hypotheses.
Perhaps if we can build on interpretability of vision and language models in domains we understand, we can eventually generalize this to other domains, but I don't think that clearly follows from anything I am reading here.
An employee at one of the worst companies in human history (META)--whose modus operandi is to destroy the social commons--complains that another of the modern era's potentially worst companies (Anthropic is, after all, offering a 10% chance of human annihilation) stole its data/code/something without proper attribution.
Rhymes with when Anthropic and OpenAI were mad that folks were distilling their models after Anthropic and OpenAI violated every intellectual property law they possibly could to build said models.
(I left Meta in September 2022, so I haven't worked there in four years. My current coworkers can attest to the fact that I'm not proud of working at Meta, but I am proud of the world-class nature of what our teams built. In any case, I don't believe my personal feelings about Meta have any bearing on the matter.)
I mean they were evil from the beginning; my point is that she has has had plenty of time working at a non-evil company to not be dragged down for it. One could serve less time in prison for armed bank robbery
Often, I see 50 people all recording the same exact thing on their iphones. Rather than a single 200MB video, we create 10GB of data. Then I think about how many people are paying a subscription to Apple for icloud storage.
The amount of photo storage being used is impressive. Automatic backup to icloud and server provisioning to reliably store it all is also impressive. I just wonder if it makes sense.
Frenetic economic activity in service of other frenetic economic activity cannot, by itself, be taken as a reflection of quality-of-life improving growth.
Suppose that subsidies lead to the price of Ubers falling close to zero, with drivers also being incentivized to drive more to support the (artificial) increase in demand -- artificial because it is driven by subsidies funded by new money creation. Traffic increases and in response, new roads are built.
reply