This is my question too. If we are all just going “well that sucks” after AI solves this problem, why did anyone care about the problem being solved in the first place?
Is the bummer that we got a solution we didn’t want - that navier-stokes is not always applicable or something, but we hoped it was?
I think the Navier Stokes problem kind of illustrates what he’s highlighting. I think most people even before AI expected that this would resolve in the negative and that you could get finite time blow up. There wasn’t really ever going to be a situation where the resolution to this question, or really any of the other Millenium Prize problems as far as I know, gives some kind of immediate massive practical feedback.
The hope with many of these problems in math is that in trying to prove that, we get some additional insight into why it blew up that could be applied elsewhere to more general PDEs that cannot be easily controlled.
I think the observation from Tao and many others is that when humans solved these problems, the additional insights into intuition and theory building came for free since humans can give expository on what they found hard or what was their own intuition. This is much more difficult or tedious to extract from an AI model. Even when people did have access to the chain of thought, it wasn’t always very helpful to figure out what was the exact thing that made it all click. This is even more difficult how that the CoT are hidden but I would think the sort of difficulty of extracting the key ideas for a human might be worse now with more advanced models.
There’s a long term aspect to this too where we have historically used these problems as markers for the other parts of mathematics but if AI can solve it all, then suddenly this signal is not very meaningful.
Maybe to bring it closer to home. If an oracle just gave you P \neq NP, then this would be generally uninteresting since this was already expected. There’s a deeper question of why that needs to be answered. However, one would hope that creating such a separation would give us tools that allow us to create lower bounds on a lot more problems we do care about and perhaps some bigger insight onto what makes a problem intrinsically hard or easy. These long term considerations are helpful but are definitely more vague. The remarkable part is that AI is separating the part about proving theorems and the “free” insight you get.
Because the mathematicians consulted 25 or so years ago believed their solutions would lead to the greatest amount of interesting new maths to explore, and because they had been validated as being hard by being attempted and not solved for a long time.
Honestly, the attitude of the math community is a bit cringe and increasingly I think some of the elite/mystical aura is fading. Rather than a rich fertile jungle where AI can barely chomp through a fraction of the luscious terrain, one gets the sense it's a desert and all the oases are running dry.
I use a Behringer X32 rack with a couple digital snakes to route audio between 20 or so radios, software defined receivers, online broadcasts and other audio sources, and it's one of the most powerful pieces of technology in my lab/office.
The quality however of the X32 leaves a little bit to be desired. The power supply died in it due to fault capacitors (I was able to replace them myself) and the rotary switches on the rack unit itself no longer work (I remote control the box anyway) but.. I've been able to look past all this because the sheer power of the tech is unbeatable for my use case. Running a single CAT-5 connection from my detached lab/rack to my office in the house with 32 channels of AES audio is amazing stuff for me.
I agree. If this was Oracle I might not have too difficult of a time believing this is all of the story. But I do think in this case there is more to the story.
I'm not sure I'd call this risk. Risk would be "you can invest the money, but you might not get it back" however the above is referring to the "a 51% attack absolutely works but you need a shit ton of money to do it" aspect instead. This makes it capital intensive, not (necessarily) risky.
The fact that it succeeds does not mean that you get the money back (eg the price of monero could drop if that happens). You may also have miscalculated some parameters in all this or something unexpected happens (where human factor is involved). So there should always be risk involved imo. Otherwise I agree, even in a probability 1 success situation this would still not be called "cheap".
It is absolutely risky. Your facilities can burn down once the ASICs arrive and before they are turned on, or your employees simply steal them for their own uses. Heck, you can have a fire once they get powered-on, because a power cable was poorly made. You might get sent the wrong product, or you could be ghosted without a delivery.
Expensive is a better fit than capital intensive, because there are massive ongoing costs to actually perform the attack, electricity for one.
If you want to understand the risks for a project, pretend you are at arms length and are being asked to fund the project 100% up-front. You'll find a huge list of risks very soon.
This is why I didn't say it made the investment risk free, I said being capital intensive does not make something (inherently) risky. There is no such thing as an investment without risk, but how risky it is is largely orthogonal to how capital intensive it is, and the above was talking about the latter so using the term "risk" for that half is not a great correction.
Having the power to deny others to mine blocks does not mean that you can obtain the tokens from their wallets. Miners can't sign transactions on users' behalf. You can rewrite all of history but then no exchange will accept your version of it to let you exchange the tokens for fiat. Also this will almost certainly crash the price of XMR substantially. And later people will be able to fork/restore the original version. The technological side of the blockchain is only part of the consensus/trust/market/popularity. People are the other part, and people will not pay the attacker for their successful attack.
The attacker doesn't need to steal tokens. They just need to short the token while they sufficiently disrupt the network to drive down the price. They get the money and your tokens become worthless.
I was completely wrong about the cost. XMR mining rewards amount to only $150k/day.
At the height of the attack, Qubic (the company) paid people up to $3 in QUBIC for every $1 of XMR they mined through QUBIC, and they achieved around 33% of XMR's hashrate which was sufficient to mine the majority of blocks for a few hours.
If they were forced to buy back all those QUBICs they paid out, this might have cost them ~$100k/day. But thanks to the media attention it's likely that they didn't need to buy anything back and actually were able to emit more than they otherwise could have.
XMR needs to adapt -- switch to PoS, or ASICs-based POW, or a hybrid of both.
Controlling 51% of XMR costs ~$30M per day, you'd have to short a huge amount of XMR to make that worthwhile. Who would be the counter party and how would you do that anonymously?
The attack itself is unprofitable, the "profit" for Qubic is the publicity they get. (or at least that's what they're betting on)
Monero has a theoretical market cap of $4.7B USD and daily volumes >$100M USD. I wouldn't recommend taking that short position in one go but over a few days and a few exchanges I wouldn't see a problem acquiring a very large short of the token.
Ask yourself why so many content hosting platforms utilize CLoudflare's services and then contrast that perspective with your posted one. Might enlighten you a bit to think about that for a second.
Between this and Hertz's new AI damage detection models, we're seeing the enshitification of business travel reaching a new level, and also doing a great job of really ticking off a group of customers (business travelers) who are already irritated enough.
Rest markets itself as a way to "unlock a new revenue stream"
Leave it to the bean counters to see this as an opportunity to generate new revenue streams from customers while simultaneously pissing them off.
There have always been attempts to launder fraud through intermediaries - computerized, bureaucratic, or otherwise. They think (well, know) if they abstract or obfuscate things in a novel way, that they'll have enough time to hit markets across states without sophisticated legislation before the legal immune systems can respond, potentially years later.
This type of algorithmic grift is transparent to judges and people with common sense, but there doesn't seem to be a lot of interest at or outside of the federal level through regulators like the FTC to prevent it, just curtail certain circumstances.
That sounds like stuff of science fiction, can't believe it works. The best part is that it works long distance without having to have satellites in the sky... and is probably un-jammable?
Thanks for sharing this, so cool to learn about it!
