1. Introduction: <https://news.ycombinator.com/item?id=47689648> (619 comments)
2. Dynamics: <https://news.ycombinator.com/item?id=47693678> (0 comments)
3. Culture: <https://news.ycombinator.com/item?id=47703528>
4. Information Ecology: <https://news.ycombinator.com/item?id=47718502> (106 comments)
5. Annoyances: <https://news.ycombinator.com/item?id=47730981> (171 comments)
6. Psychological Hazards: <https://news.ycombinator.com/item?id=47747936> (0 comments)
And this submission makes:
7. Safety: <https://news.ycombinator.com/item?id=47754379> (89 comments, presently).
There's also a comprehensive PDF version for those who prefer that kind of thing: <https://aphyr.com/data/posts/411/the-future-of-everything-is...> (PDF) 26 pp.
(Derived from aphyr's comment: <https://news.ycombinator.com/item?id=47754834>.)
In what world would I ever expect a commercial (or governmental) entity to have precise alignment with me personally, or even with my own business? I argue those relationships are necessarily adversarial, and trusting anyone else to align their "AI" tool to my goals, needs, and/or desires is a recipe for having my livelihood completely reassigned into someone else's wallet.
I guess I'm trying to wonder why this line of thinking (in theory) doesn't turn to paranoia about everybody. I don't know much ethics or political theory or anything.
It does. People drive these entities. People hide behind the liability shields and authority of these entities. Also notice that I generalized with the phrase “…and trusting anyone…”
It really isn't. The whole point of the market system is to collectively align people's actions towards a shared target of "Pareto-optimized total welfare". And even then the alignment is approximate and heavily constrained due to a combination of transaction costs (which also account for e.g. externalities) and information asymmetries. But transaction costs and information asymmetries apply to any system of alignment, including non-market ones. The market (augmented with some pre-determined legal assignment of property rights, potentially including quite complex bundles of rules and regulations) is still your best bet.
Uh, what? People have been killing each other over values misalignments since there have been people. We invented civilization in part to protect our farms and granaries from people who disagreed with us on whose grain was in said granaries.
Not OP, but for me, kind family and friends, and various feel-good pieces of fiction and other writing, at least let me envision the possibility of a perfectly kind/dedicated/innocent/naieve individual who is truly on my side 100%. But even that is mostly imagination and fiction... although convincing others of that isn't necessairly an argument worth making.
Commercial entities have a fundamental purpouse of profit. While profit doesn't have to be a zero-sum game - ideally, everyone benefits in a somewhat balanced way - there's some fundamental tension, in that each party's profit is necessairly limited by the other party's.
Government entities have a fundamental purpouse of executing the will of the state, which is rather explicitly not the same thing as the will of you as an individual.
Both commercial and government entities also tend to involve multiple people, which gets statistics working against you - you really gathered that many people who would put your needs above their own, with exactly zero "imposters" - which in this context just means people with a bit of rational self interest?
> I guess I'm trying to wonder why this line of thinking (in theory) doesn't turn to paranoia about everybody. I don't know much ethics or political theory or anything.
Just because you're paranoid, doesn't mean they aren't out to get you. Trust, but verify.
You might not be able to put absolute blind trust in anybody. I certainly can't. However, one can hedge one's bets, and diversify trust. Build social circles of people with good character, good judgement, and calm temperments - and statistics will start working for you. It's unlikely they'll all conspire to betray you simultaniously, especially if you've ensured betrayal costs much and gains little. While petty and jealous people can indeed be irrational enough to betray under such circumstances, it'll be harder for them to create the kind of conspiracy necessary for mass betrayal that might cause significant enough damage to warrant proper paranoia. You might still have to watch out for gaslighters stealing credit (document your work!) and framing people (document your character!) and other such dishonest and manipulative behavior... but if everyone's looking out for the same thing, well, that's just everyone looking out for everyone else! That's a community looking out for each other, and holding everyone honest and accountable. Most find comfort in that, rather than the stress paranoia implies.
Put yourself in a room full of manipulators and schemers, on the other hand, and "parnoia about everyone" might be the only reasonable or rational response!
Profit is obtained by maximizing traded benefits and minimizing costs. None of this requires taking anything away from any other party.
There was a Japanese visual novel in the 2000s about a girl who was your personal maid, and was so devoted would always take your side in any conflict, accept and support you just the way you are, even if you were a horrid person to your friends. It turns out she was a ghost, or a kind of yokai, or sonething. Anyhoo, back on 2ch she attracted a fandom, and there was a second group of people on 2ch who labelled her a "useless person manufacturer" because if you actually had a person who always accepted you just the way you are and never pushed back, that can be actually a trap that prevents you from developing.
It puts even certain utopian AI fiction, like Richard Stallman's story "Made for You", into a whole new light.
Seems like a strawman, I don't think anyone means this when talking about alignment.
More general goals, like avoiding paperclip maximization, are broadly applicable to humanity.
This is true, and I believe that the "sufficient funds" threshold will keep dropping too. It's a relief more than a concern, because I don't trust that big models from American or Chinese labs will always be aligned with what I need. There are probably a lot of people in the world whose interests are not especially aligned with the interests of the current AI research leaders.
"Don't turn the visible universe into paperclips" is a practically universal "good alignment" but the models we have can't do that anyhow. The actual refusal-guards that frontier models come with are a lot more culturally/historically contingent and less universal. Lumping them all under "safety" presupposes the outcome of a debate that has been philosophically unresolved forever. If we get hundreds of strong models from different groups all over the world, I think that it will improve the net utility of AI and disarm the possibility of one lab or a small cartel using it to control the rest of us.
Most countries have a pretty strong ban on most kinds of weapons, the US is one of the few that lets everyone run around with their rooty tooty point and shooty, but most countries have implemented bans. Some because the government doesn't want the people having them, and in others the citizens call for the bans because they don't like the idea of getting shot by their fellow citizens.
It won't be long before citizens and governments get tired of models being used for criminal activities and will eventually lay down laws around this. Models will have to be registered and safety tested, strict criminal prosecution will happen if you don't. And the big model companies will back their favorite politicians to ensure this will happen to.
Now, that in general will be helpful as there will still be more models, but it will still not be a free for all.
Anyone outside the UK can share what this is about?
Alignment is a Joke Well-meaning people are trying very hard to ensure LLMs are friendly to humans. This undertaking is called alignment. I don’t think it’s going to work.
First, ML models are a giant pile of linear algebra. Unlike human brains, which are biologically predisposed to acquire prosocial behavior, there is nothing intrinsic in the mathematics or hardware that ensures models are nice. Instead, alignment is purely a product of the corpus and training process: OpenAI has enormous teams of people who spend time talking to LLMs, evaluating what they say, and adjusting weights to make them nice. They also build secondary LLMs which double-check that the core LLM is not telling people how to build pipe bombs. Both of these things are optional and expensive. All it takes to get an unaligned model is for an unscrupulous entity to train one and not do that work—or to do it poorly.
I see four moats that could prevent this from happening.
First, training and inference hardware could be difficult to access. This clearly won’t last. The entire tech industry is gearing up to produce ML hardware and building datacenters at an incredible clip. Microsoft, Oracle, and Amazon are tripping over themselves to rent training clusters to anyone who asks, and economies of scale are rapidly lowering costs.
Second, the mathematics and software that go into the training and inference process could be kept secret. The math is all published, so that’s not going to stop anyone. The software generally remains secret sauce, but I don’t think that will hold for long. There are a lot of people working at frontier labs; those people will move to other jobs and their expertise will gradually become common knowledge. I would be shocked if state actors were not trying to exfiltrate data from OpenAI et al. like Saudi Arabia did to Twitter, or China has been doing to a good chunk of the US tech industry for the last twenty years.
Third, training corpuses could be difficult to acquire. This cat has never seen the inside of a bag. Meta trained their LLM by torrenting pirated books and scraping the Internet. Both of these things are easy to do. There are whole companies which offer web scraping as a service; they spread requests across vast arrays of residential proxies to make it difficult to identify and block.
Fourth, there’s the small armies of contractors who do the work of judging LLM responses during the reinforcement learning process; as the quip goes, “AI” stands for African Intelligence. This takes money to do yourself, but it is possible to piggyback off the work of others by training your model off another model’s outputs. OpenAI thinks Deepseek did exactly that.
In short, the ML industry is creating the conditions under which anyone with sufficient funds can train an unaligned model. Rather than raise the bar against malicious AI, ML companies have lowered it.
To make matters worse, the current efforts at alignment don’t seem to be working all that well. LLMs are complex chaotic systems, and we don’t really understand how they work or how to make them safe. Even after shoveling piles of money and gobstoppingly smart engineers at the problem for years, supposedly aligned LLMs keep sexting kids, obliteration attacks can convince models to generate images of violence, and anyone can go and download “uncensored” versions of models. Of course alignment prevents many terrible things from happening, but models are run many times, so there are many chances for the safeguards to fail. Alignment which prevents 99% of hate speech still generates an awful lot of hate speech. The LLM only has to give usable instructions for making a bioweapon once.
We should assume that any “friendly” model built will have an equivalently powerful “evil” version in a few years. If you do not want the evil version to exist, you should not build the friendly one! You should definitely not reorient a good chunk of the US economy toward making evil models easier to train. ...
You don't need to train new models. Every single frontier model is susceptible to the same jailbreaks they were 3 years ago.
Only now, an agent reading the CEOs email is much more dangerous because it is more capable than it was 3 years ago.
The cynic in me agrees with the article’s premise, but not because I believe "alignment is a joke", but because I doubt that humans are "biologically predisposed to acquire prosocial behavior."
Seems easy enough, I'm actually pretty confident in even the most incompetent of current world leaders in this particular task.
Geoffrey Hinton will not have his liver pecked out every day like Prometheus does.
I think the author is brushing against some larger system issues that are already in motion, and that the way AI is being rolled out are exacerbating, as opposed to a root cause of.
There's a felony fraudster running the executive branch of the US, and it takes a lot of political resources to get someone elected president.
I'm seeing that these tools are extremely powerful the hands of experts that already understand software engineering, security, observability, and system reliability / safety.
And extremely dangerous in the hands of people that don't understand any of this.
Perhaps reality of economics and safety will kick in, and inexperienced people will stop making expensive and dangerous mistakes.
”The future” is happening because it is allowed in our current legal framework and because investors want to make it happen. It is not ”happening” because it is good or desirable or unavoidable.
How did brains acquire this predisposition if there is nothing intrinsic in the mathematics or hardware? The answer is "through evolution" which is just an alternative optimization procedure.
This "just" is... not-incorrect, but also not really actionable/relevant.
1. LLMs aren't a fully genetic algorithm exploring the space of all possible "neuron" architectures. The "social" capabilities we want may not be possible to acquire through the weight-based stuff going on now.
2. In biological life, a big part of that is detecting "thing like me", for finding a mate, kin-selection, etc. We do not want our LLM-driven systems to discriminate against actual humans in favor of similar systems. (In practice, this problem already exists.)
3. The humans involved making/selling them will never spend the necessary money to do it.
4. Even with investment, the number of iterations and years involved to get the same "optimization" result may be excessive.
(Apart from that, I'm generally suspect of evolution-based arguments because they are often structurally identical to saying “God willed it, so it must true”.)
Large language models are not evolving in nature under natural selection. They are evolving under unnatural selection and not optimizing for human survival.
They are also not human.
Tigers, hippos and SARS-CoV-2 also developed ”through evolution”. That does not make them safe to work around.
while under the umbrella of evolution, if you really want to boil it down to an optimization procedure then at the very least you need to accurately model human emotion, which is wildly inconsistent, and our selection bias for mating. If you can do that, then you might as well go take-over the online dating market
Putting aside malicious actors, the analogy here means benevolent actors could spend more time and money training AI models to behave pro-socially than than evolutionary pressures put on humanity. After all, they control the that optimization procedure! So we shouldn’t be able to point to examples of frontier models engaging in malicious behavior, right?
Such a fear mongering position. You can learn to build pipe bombs already. Take any chemical reaction that produces gas and heat and contain it. Congratulations, you have a pipe bomb.
Meanwhile.. just.. ask an LLM if you can mix certain cleaning chemicals safely.
> I see four moats that could prevent this from happening.
Really? Because you just said:
> human brains, which are biologically predisposed to acquire prosocial behavior
You think you're going to constrain _human_ behavior by twiddling with the language models? This is foolishly naive to an extreme.
If you put basic and well understood human considerations before corporate ones then reality is far easier to predict.
the cost of the wrong answer to this question is so incredibly high that I hope nobody is sincerely asking an LLM for this information. The things people trust to "machine that gives convincing answers that are correct 90% of the time" continue to shock me
The internet produced 4chan. Produced scammers. Produced fraud. Instrumental in spreading child porn. Caused suicides. Many people lost their lives due to bullying on the internet. Many develop have addictions to gaming.
To anyone who has given it some thought, any sufficiently advanced technology usually affects both in good and bad ways. Its obvious that something that increases degrees of freedom in one direction will do so in others. Humans come in and align it.
There's some social credit to gain by being cynical and by signalling this cynicism. In the current social dynamics - being cynical gives you an edge and makes you look savvy. The optimistic appear naive but the pessimists appear as if they truly understand the situation. But the optimists are usually correct in hindsight.
We know how the internet turned out despite pessimists flagging potential problems with it. I know how AI will turn out. These kind of articles will be a dime a dozen and we will look at it the same way as we look at now at bygone internet-pessimists.
This is response not just to this article, but a few others.
If the AI tech keeps going at the direction it's going now, more and more people will start believing the world would be better if the internet and computer had never been invented.
You talk like the internet being a net positive is a given. It really isn't, especially after it's proven that it doesn't democratize power (see Arab Spring, and China, and the US, and everywhere.)
Ask any poor person in India what their sentiment is with tech - it is usually optimism.
> You talk like the internet being a net positive is a given. It really isn't, especially after it's proven that it doesn't democratize power (see Arab Spring, and China, and the US, and everywhere.)
The world is far more democratic now than before and I attribute it to technology because it reduces information asymmetry.
That is fantasy. Information technology has created an unprecedented level of information asymmetry and the gap is widening everyday as the total computing capacity grows.
Before information era, the ruling class was roughly as blind as peasants. Population census took years, and sometimes outright impossible. The opaqueness was two-way. Now it's one way - people in power know everything about the citizens.
(hint: there already exist examples like such)
Without information, there is no way a voter may know which person to vote for and whether to believe in them at all and you are easily susceptible towards manipulation.
It will become more clear when you try to answer this hypothetical: if your objective were to bring in more democracy in North Korea, would you allow the global internet to proliferate if you could? According to your theory, it would just make it worse in general.
A sludge of spyware and addiction machines which employ negative emotion and outrage to drive shareholder value?
"The internet" is a pretty big tent. Everything from text messages to streaming video to online gaming to social media to encyclopedias. I think 15 years ago you could make a strong case that the internet was mostly a net positive, I think now that is much more difficult. If governments are able to fully realise their plans for surveillance and control, it will almost certainly become a net negative. Of course with many positive aspects.
So likewise with AI, we should be careful to not make the same mistakes as we did with the internet so we can realise something that is mostly positive. We could absolutely have a world where AI is as beneficial as you believe it will be, but we don't get there through inaction, we get there by being deeply critical of the negative aspects of AI and ensuring that we don't let a small number of hyper scalers control our access to it.
I do think that safety is important. I'm particularly concerned about vulnerable people and sycophantic behavior. But I think it's better not to be a luddite. I will give a positively biased view because the article already presents a strongly negative stance. Two remarks:
> Alignment is a Joke
True, but for a different reason. Modern LLMs clearly don't have a strong sense of direction or intrinsic goals. That's perfect for what we need to do with them! But when a group of people aligns one to their own interest, they may imprint a stance which other groups may not like (which this article confusingly calls "unaligned model", even though it's perfectly aligned with its creators' intent). People unaligned with your values have always existed and will always exist. This is just another tool they can use. If they're truly against you, they'll develop it whether you want it or not. I guess I'm in the camp of people that have decided that those harmful capabilities are inevitable, as the article directly addresses.
> LLMs change the cost balance for malicious attackers, enabling new scales of sophisticated, targeted security attacks, fraud, and harassment. Models can produce text and imagery that is difficult for humans to bear; I expect an increased burden to fall on moderators.
What about the new scales of sophisticated defenses that they will enable? And for a simple solution to avoid the produced text and imagery: don't go online so much? We already all sort of agree that social media is bad for society. If we make it completely unusable, I think we will all have to gain for it. If digital stops having any value, perhaps we'll finally go back to valuing local communities and offline hobbies for children. What if this is our wakeup call?
[1] https://en.wikipedia.org/wiki/Wikipedia:Signs_of_AI_writing
For the future, try to avoid prevaricating when you actually have a clear sense of what you want to argue. Instead of convincing me that you've weighed both options and found luddism wanting, you just come off as dishonest. If you think stridently, write stridently.
I may think stridently (debatable) but I generally believe it is best to always try to meet in the middle if the goal is genuine discussion. This is my attempt at that.
> The issue with most of these articles is that they seem to demonize the technology, and systematically use demeaning language about all of its facets.
This is very confident, strident language. You clearly believe that there is a faction of people demonizing technology, akin to luddites, who are not worthy of being taken seriously.
> This one raises a lot of important points about LLMs, but...
So here you go for the rhetorical device of weighing the opposing view. Except, you don't weight it at all. You are not at all specific about what those points are. It's just a way to signal that you're being thoughtful without having to actually engage with the opposing viewpoint.
> I do think that safety is important... But I think it's better not to be a luddite.
Again, the rhetoric of moderation but not at all moderate in content.
It was a clear mistake to think that this was LLM writing. But I suspect the reason I made this mistake is that AI writing influences people to mimic surface level aspects of its style. AI writing tends to actually do the "You might say A is true, but B has some valid points, however A is ultimately correct." Your writing seems like that if you aren't reading it closely, but underneath that is a very human self-assuredness with a thin veneer of charitability.
I think the point was never to bring a solution or show any essence of reality. The point was being polemical and signalling savviness through cynicism.
On a serious note, I think they meant TN, as in Torment Nexus, but I could be wrong.
1. Introduction: 33,088 (https://news.ycombinator.com/item?id=47689648)
2. Dynamics: 3,659 (https://news.ycombinator.com/item?id=47693678)
3. Culture: 5,914 (https://news.ycombinator.com/item?id=47703528)
4. Information Ecology: 777 (https://news.ycombinator.com/item?id=47718502)
5. Annoyances: 7,020 (https://news.ycombinator.com/item?id=47730981)
6. Psychological Hazards: 199 (https://news.ycombinator.com/item?id=47747936)
Feedback from early readers was that the work was too large to digest in a single reading, so I split it up into a series of posts. I'm not entirely sure this was the right call; the sections I thought were the most interesting seem to have gotten much less attention than the introductory preliminaries.
I had never heard of you, and this article appeared very biased to me. I found the information ecology piece superior, shame that it went unnoticed; I will try to go through all of them. I admire the breadth of topics you’re covering and appreciate the many sources. They’re clearly written in your own voice and that is great to see, I guess I mostly reacted to not being fully aligned with your view.
I suspect that if you'd not broken up the post into a series of smaller ones, the sorts of folks who are unwilling to read the whole thing as you post it section by section would have fed the entire post to an LLM to "summarize".
I still remember his takedown of mongodb's claims with the call me maybe post years and years ago filling me with a good bit of awe.
If ‘tptacek posts a blog post, I bet it similarly does well, on average, because they’re a “known quantity” around these parts, for example.
Also, if you think this is just “LLM is bad”, I highly suggest reading the series first. The social impacts they talked about at the start of the series should resonate with a lot of people here and are exactly the kind of thing which people building systems should talk about. If you’re selling LLMs, you still want to think about how what you’re building will affect the larger society you live in and the ways that could go wrong—even if we posit sociopath/MBA-levels of disregard for impacts on other people, you still want to think about how LLMs change the fraud and security landscape, how the tools you build can be misused, how all of this is likely to lead to regulatory changes.
1. AI becomes a highly protected technology, a totalitarian world government retains a monopoly on its powers and enforces use, and offers it to those with preexisting connections: permanent underclass outcome
2. Somehow the world agrees to stop building AI and keep tech in many fields at a permanent pre-2026 level: soft butlerian jihad
3. Futurama: somehow we get ASI and a magical balance of weirdness and dance of continual disruption keeps apocalypse in check and we accept a constant steady-state transformation without paperclipocalypse
Its like listening to Christians talking about the rapture.
Frontier AI models get smarter every year, humans stay relatively just as smart. If you don't believe that somehow AI will just suddenly stop getting better (which is as much a faith-based gamble as assuming some rapturous outcome for AI by default), then you'd have to assume that at some point AI will surpass human intelligence in all fields, and the keep going. In that case human minds and overall will will be onconsequential compared to that of AI.