moktonar

The hack is actively run by an entity. The virus is completely autonomous and replicating. Add in the capability of adding zero day capabilities to it, polymorphism, etc and your pretty much fucked

The hacks can be stopped, the virus can be a whole another level

Air gapped networks might push the models to write computer viruses in order to bridge the gap. This is potentially very dangerous IMHO

I bet if we live in a simulation then P == BQP, and viceversa. I’d love to throw this at those models..

Finally I start to see what will save the situation: you can only advance AI up to your highest level of intelligence, after that you start losing control and thus cannot reliably advance. That’s the mechanism that will make AI development viable. It won’t be improving exponentially, instead it will improve at the rate our intelligence can. This is both reassuring and a very good example of how co-evolution works. Expect an s-curve soon, but also this is the moment where we as humanity will start benefiting.

Ethics matter a lot, you can do whatever you want and be considered a very bad human being as a result. The result matters, but it also matters the discrepancy between what you do and what you claim you do. Bruteforce != ai did it. And that’s fact, no matter how much river you cry me

Shouldn’t this be a democratic process?

Ruining a lifetime worth of research effort, passing bruteforce as achievement, thus lying, just to appear first in position, is something to be ashamed of IMO.

They took a lifetime of research effort, threw millions of dollars of compute at it and claimed their AI solved a MPP, and they did all this just to be the first to claim the prize. Once you knew the direction, given the new insights on the subject, with all that compute you could probably have just bruteforced your way to a solution, and thousands of agents are not the same thing as “our model came out with a solution on its own”. This is pure greediness, and require at least some apologies to the involved researchers and to the whole community for showing such bad ethics and behaviour

OpenAI should just apologize for having been too greedy. That’s it, as simple as it gets. The fact that it never will is the biggest red flag.

As I said: it never really was. And another important thing to put forward is: it’s not the AI the “zero days everything” the bugs are already there, just unexplored.. it’s just surfacing the reality

The real bigger elephant in the room is: assume nothing is safe anymore (not that it ever was, but now more than ever)

The comments problem is a big one and no matter how hard I try the LLM will always write them verbosely. I find that including the rules in each prompt is fare more effective than once in the context

Surprise! The man in the middle man-in-the-middles! This is only the beginning, when you’ll get used to this they’ll do worse and worse, enshittification, remember?

They should try to forecast earthquakes, that would really be a breakthrough If anything better than random comes out

Waiting for the “Our model inadvertently hacked the nukes and launched Armageddon” title

The problem is that the more AI eats labor, the more you can hike the costs until you pretty much can match the salary of the workers you replaced, with some margin, enough for the user to accept the cost. That’s what will happen in the next decade IMHO. Real cost = base expense + what user accepts to pay

Did the Feds desperately need a way of getting the key? is this a bugdoor? Has the commits been traced? Recently I’ve been seeing this pattern a lot and I’m starting to be a little bit suspicious. Maybe it’s because people are more sensible to this and post more on it?

And so it begun..

Jokes aside does anybody else notice the eerie similarities with gray aliens?

I think we should start having ai versions like beta ora alpha versions and then consolidate them into human made versions with time, at least one is free to stay safe or on the bleeding edge as one likes and we all get a win-win best-of-all-worlds situation (hopefully)

Inexistent security, absent security contacts/hard to get in touch with, denial/delay/won’t patch, most functionality to deploy a backdoor is already present, to me equals bugdoor. This is wanted behavior, not an accident, and is a widespread pattern..

int x[n] and int *x are very different things when it comes to defining memory layout tho. In one case you end up with n int sized slots of memory, in the second with one register sized slot. That makes all the difference when defining structs for example.

I get what he’s saying, but, doesn’t he compare classical speed up of parallelizing 64 bit key space on 2^16 cpus with parallelizing 128 bits key space on QCs? It’s true that sqrt (2^128/2^16) = 2^56 and that 56 >> 48, but in one case you are attacking a 64 bits key space and in the other a 128! If you parallelize 2^128 on 2^16 CPUs you get 128-16=112 bits of key space per cpu which is much bigger than 56! No?

Edit: I mean, I get the point is to prove that 2^128 on QC is not the same as 2^64 on CC but it’s still a lot less to search. If a paper came out with that big of a key space reduction AES would be considered broken IMO

Except maybe storing another smaller vector for the difference with the original data an also quantize that maybe recursively

Aren’t polar coordinates still n-1 + 1 for radius for n-dim vector? If so I understand that angles can be quantized better but when radius r is big the error is large for highly quantized angles right? What am I missing?

Is it possible to do the inverse, then? (Tranforming weights back to code)

Thank you for all the work guys, I’ll see how I can help.

Well fucking done. Anthropic has just gained the “has bollocks” status. Also now we know what the govt is really up to with AI. G fucking g

The random uuid selection is far superior because of lifespan, you can only have so many functioning devices at the same time, and on the contrary to tree-based uuids once a device is decommissioned the uuid can be reclaimed. Practically though it would probably be a mixed algorithm where positioning would give the id root and the rest is selected randomly

UB is the definition of Free Will that’s why you can’t control it, and for a programmer something that cannot be controlled is felt as dangerous..