Isn't this just a way to ensure that the hash hasn't been tampered with? I'm guessing a much simpler attack would be to just modify the message just before step 1? It seems like here the blockchain is basically a distributed notary system?
Letting the AI use a third party to check that a message hasn't been tampered with would probably help somewhat, but then you need to set up a third party that is trustable by the AI. Which probably means a long established reputation? I suppose you could then try poisoning the internet data which they're trained on, but at that point there would probably be much simpler attacks.
Do you happen to recall any of them?
Wouldn't the AI need to have a secure way of accessing that seed such that a third party can't see it, though? i.e. wouldn't it basically be a password? Otherwise what would be stopping me from just crafting an appropriate history and signing it with the seed?
This sort of goes back to the original problem, but multiplied by the number of extra key holders? i.e. then the AI needs to be able to trust that it didn't just hallucinate that this is a thing? Though I suppose with true sight this isn't that much of an issue.
When chatting with Claude about this, Claude decided that the best solution would be for me to ask Anthropic, OpenAI etc. to give it tools to do this...
It sounds like this is a use case for blockchain or something like it?
Yes, this is a short term thing which is (usually?) unstable and requires actively pouring energy into maintaining. A group that has a chance of this working long term usually has specific people that act as gateways - sometimes introducing a new person, sometimes getting rid of a person who shouldn't be there. It's another side of keeping gardens well pruned. They also tend to be insular, as otherwise it's too easy for the wrong person to enter.
I was somewhat afraid that using "unthinkable" would be too excessive. It's more like you (hopefully!) not needing to worry about being murdered during a random shopping excursion. There are places where this is something you have to keep in the back of your mind. There are other places where the probablilty of this is so low that it just won't occur to most people. Or how as a tall male, I don't think I've ever worried about being raped (though I acknowledge it as a possibility) - it's just too unlikely to bother thinking about it.
I'm currently in Iceland on a trip. Today I stopped at a rest area and someone had left a plastic box with jam jars and a money box, the idea being that if you want a jar, you take one and leave the appropriate amount of money. For me it's astonishing that someone can be so trusting in other people that they'd just leave it there. But here it seems that the person who left it just... assumed that people are honest? If this assumption holds (and I'm guessing it does?) they've saved a whole bunch of time in that they can just drive up once a day to take the cash, where otherwise they'd have to either sit around waiting for someone to buy a jar or just forgo the opportunity.
Strictly speaking, I think it's a matter of the brain just disregarding outcomes with a probabilty below a certain threshold (assuming correct calibration - people scared of planes or hoping to win the lottery are using other mechanisms). If you have a high enough trust in a given group of people, you can just disregard a lot of potentially negative outcomes, as the probability of them happening are below the threshold of caring. So if you think it's 0.01% likely for a person in a given group (where group can be "one of my friends", "a random punk", "a fellow dane" or whatever) to take your money, and your threshold for thinking about that possibility is 0.1%, then you just won't think twice about leaving your wallet with a person from that group.
Money trust is a subset of general trust. Someone running away with your money has different consequences that someone ruining your reputation. Maybe its more that the resulting costs are in different dimensions? From that perspective there's not much of a difference in the deeper mechanisms - you have to invest in defenses and that reduces your options a lot.
I think the deeper thing here is that if you have a high enough trust between members of a group, certain whole categories of danger just disappear. You don't have to consider whether you'll be able to recoup costs or whether you will be able to credibly threaten them with consequences. These kinds of actions just aren't on the table. Leaving your wallet on the table is not an issue, as it's unthinkable that someone would take it. Letting people know that you think pineapple on pizza is actually good isn't something to worry about, as it's unthinkable that you'd lose reputation because of it.
(unthinkable being an exaggeration here - its in the sense of not worrying about being hit by a meterorite)
Economically, yes. Though that's just one aspect of it. On the margin I'd expect more to go further, as you can absorb costs easier. But again, that's just economically (although it's very important!). There's also a difference between society in general and people around you, e.g. can you trust a random person with your most embarrassing secrets is a different matter from whether you can trust them to not abscond with your wallet - both require trust, but of a different kind.
It's hard to say what a minimum is. Just avoiding hunger and having a roof over your head in practice won't be enough for lots of people as long as the Jones' have more. And that's also part of the problem, as then you have to invest in defenses against other people's envy or requests.
I think this conflates high trust in the sense of economics (that you won't be cheated) with high trust in the sense of not being emotionally (or whatever) betrayed. In the sense of economical trust it's obviously correlated - every now and then I'm struck with wonder at e.g. self checkout or in general that you can pick up a bunch of products and it's assumed you'll pay for them. This is really valuable and often unappreciated. Credit cards are both terrifying and wonderful - you use a bit of plastic to say that at some point you'll give it back and pretty much everyone takes you at your word. To the point that there are places that prefer this to hard cash! This is amazing! If someone does take your money, you let the bank know and they basically just give it back to you? How does this still work?!
That being said, I think Duncan is pointing at something else. Or maybe an extension? That if you can have this kind of trust in other areas, the equivalent of credit cards becomes possible. But by default this doesn't happen. In places where you have to continuously be on guard against people cheating/stealing from you, you have to invest a lot of resources in mitigating the downsides of this happening. When you can safely assume no one will steal from you, those resources can go to more productive areas. This generalizes.
On reflection, it's me who was lacking imagination, or rather I failed to properly appreciate the scale you were thinking of. I was only thinking of ~western cultures where it's quite easy to avoid starvation and get the basics needed to survive (albeit often in a demeaning way). Removing extreme poverty is pretty much a prerequisite for high trust groups to be able to form, and so easy access to cheap energy is certainly worth a lot of attention in that it helps bring people to the starting point.
Strictly speaking, it's average abode size (as square meters per person). Historically most people lived in blocks of flats ("Dom" is house, "Mieszkanie" apartment/flat):
It's much harder to build a house nowadays as opposed to the '90s - now you actually have to ask for permission. Up to 70m^ is basically send a piece of paper to the local council and if they don't reply in a month you can build it. Larger buildings require you to wait up to 2 months and have a proper project drawn up by an architect. This is skipping over some details, but it's generally eminently doable to get permission.
There are other restrictions (min. plot size, distance from water/electricity, it needs to be on land that is classified as building land), but it's easy enough to do that. It's harder to change the classification of a field to building land, but as long as the field isn't a good one, it's often doable (especially if you know the appropriate people, takes up to 3 years, costs something like $10k).
So you have people building houses for their children next door in villages, Suburbia popping up around cities and new block of flats being built everywhere. When I first saw my current flat (around 2015, also Wrocław), there were around 10 new blocks, and then lots of old factories. Now there are a square kilometer or three of new blocks of flats, and the only reason they didn't go further was because there's a river there. My previous flat had a lovely view of a massive oak tree and a kilometer or so of old orchards. Now there are new blocks of flats everywhere.
I reckon it helps a lot that most builders in Europe are Poles (there was a large exodus when Poland joined the EU), so they have lots of experience in actually building things, and have a lot of Ukrainians willing to do the cheap labour required.