Shrike
6hr ago
I think the Germans were censoring free speech since before Germany was unified; certainly they were doing it in the interwar period, when Hitler was censored.
This isn't a claim against the US using that for their own purposes, but the cultural or political tendency predates American involvement in German internal affairs.
Shrike
6hr ago
Applying this epistemic standard to the Greenland situation, the only thing I conclude is that he probably had some concessions so that Denmark would not simply go on record stating there is no deal. This does not mean that there is, in fact, a deal.
Denmark (and Greenland) are on the record stating that there is, in fact, a deal.
Shrike
7hr ago
I am familiar with this dynamic, but (at least from where I sit) a mostly-functional program is different from a mostly-true thesis, or what have you.
Shrike
17hr ago
I think that AI has been gravitating towards software development (perhaps to the detriment of other skills), in part because software development helps develop AI! But software development is fairly validation oriented – code runs or it doesn't. Someone who is lazy enough to have an AI write text with their name on it may also be too lazy to validate that text, or even too lazy to have another AI fact-check it.
You can say "yeah they are a poor user" and fair enough, but people will say things like "AIs have PhD level intelligence" or "AIs can act as a smart research intern" and then (in my experience) from time to time they go and make stuff up. Certainly lazy PhDs and research interns do this, all the time, but I don't think that demonstrates intelligence.
I'm not an AI-denier, I use it relatively consistently and I like it for research – but a lot of that is because Google is bad now (and their AI overview is atrocious, I've seen it present conflicting information), and you can use AI to get you links that you can check with your own eyeballs, and you can give it specific prompts that help mitigate other bad tendencies. But the fact that you have to do these things means that you cannot necessarily just expect AI cruising around on its own recognizance (or in the hands of someone who is not intelligent and/or used to using AI) to operate at that high level.
Shrike
17hr ago
There is no place for a conspiracy theory about secret stockpiles.
I don't think there are like 1,000 SM-6s tucked away somewhere but the US develops and even IOCs entire weapons systems in secret SAPs. Every so often (bin Laden, Maduro) you see a glimpse of what SOCCOM is playing around with, and SOCCOM is arguably one of the least useful organizations in the event of a peer conflict. It's true that SOCCOM gets an unusual (and some would say untoward) amount of shiny toys headed their way, but they are hardly the only US military branch to have secret weapons development in progress.
Shrike
17hr ago
Oh no, I've tried AI lots more than twice. Definitely witnessed more than one hallucination.
I'm not in software. The mistake I'm talking about wasn't made by me, so I don't have any insight into how it was made or how often the user was using it, or what model.
To give you the sort of idea of what happened without doxxing myself, imagine someone turned in their PhD dissertation and it (or parts of it) turned out to be AI written, and the AI had made stuff up.
Was he using a "cutting edge" model? No idea. I've had Opus goof stuff up for me before, so I don't necessarily trust even the higher-end models not to make mistakes from time to time.
Shrike
20hr ago
For Russian (or American) kit, I figured you would have good most-angle photos from parades, demos, and such. You're right though that these are mostly taken from the human level.
I imagine that another obvious issue distributing RFID tags to your troops is that they can be used by both sides. I realize that in theory they will only respond to coded radio signals that in theory only one side can transmit, but if you're putting them everywhere and then putting the transponder in something as failure prone as a kill-drone your enemy's going to be reverse-engineering your IFF on like day 2 of the war.
I've speculated, based on my knowledge of machine imaging, that we'll see "wartime camo" become a thing in future wars, with "peacetime camo" designed to be something maximally different from operational paint schemes to confuse ATR. Apparently the Russians are already using a variety of disruptive paint schemes in Ukraine. We can only hope it results in making razzle-dazzle great again!
Shrike
21hr ago
Very interesting! I actually thought that open-source imagery would be better resourced than that (although of course it's pretty easy to mess with the profile of a tank with a can of spray paint or a camouflage mesh, which I guess is part of the problem).
Always very heartening to have an industry professional tell me I haven't gone completely off the rails, thanks.
Shrike
22hr ago
I think it's worse. In the case I am familiar with, (sample size of two) as far as I can tell, when a guy did something on his own, he did okay. When he tried to use AI to do about the same thing, SNAP!
So either the AI screwed up big time 50% of the time, or 100% of the time. Either way, not great, and seems worse "per task" than self-driving cars.
Shrike
1d ago
I just don't think sticking Astra or Fable (as opposed to a lightweight image processor) is the solution here.
The problem with a lightweight tool is that it can be fooled by changing small details (e.g. using Ukrainian spelling on your Russian tank). But those are exactly the same sorts of things that are going to trip up Astra, too, because it is going to be instructed to use a similar decision matrix, which, again - why not just use a lightweight image processor if you are going to give them both the same targeting library.
In many ways, I think militaries would prefer a tool that isn't doing independent reasoning to strike a target. The reason for this is pretty obvious: independent reasoning historically leads to friendly fire. At least if your independent reasoner is on the bridge of a ship (and this goes for a human or a larger AI model) and it screws up, you can figure out what happened by conducting interviews and data-dumps afterwards.
If you design a missile with an independently reasoning seeker-head and it starts friendly-firing, you may have a hard time figuring out what is going on because there may be nothing left of the malfunctioning "misaligned" system to evaluate and its reasoning may be too complex to review in a timely manner. So if something happens, you might have to yoink an entire production line of weapons and/or re-train your model in the middle of a war. If a deterministic program is friendly-firing, it means:
- Blue-on-blue employment, which isn't the fault of the model
- Threat library incorrectly calibrated
- Target determination software incorrectly calibrated
All of which are probably easier to figure out and patch during an active conflict than "what is going on with Astra" - witness the difficulties the AI companies are having following chain-of-thought now, when they have total access and control over the models.
I am not saying that we will never get the fabled "self-aware weapons system" from AI, and I wouldn't be surprised if testing is done relatively soon. But in the current threat environment, it strikes me as buying a sledgehammer to kill a fly.
That's not to say that LLMs have no military value at all. I just don't think they are optimized for weapons guidance.
Shrike
1d ago
It is still a problem with whatever models people are using. Just had to deal with a pretty bad incident at work a couple of weeks ago (thankfully not one due to any failures on our part).
Shrike
1d ago
elevator operator.
Hilariously I've been in elevators with an operator.
there is a silver lining for accountants: knowing which rules to bend or break
I'm pretty sure another silver lining for accountants is that "my freakin' AI did it wrong" is probably not an excuse that the IRS will accept. They will accept "my accountant did it wrong."
Shrike
1d ago
I can only assume that you haven't of the Vogons.
Surely that sort of thing is in the domain of the CIA!
So they do have incentive to cram as big models as possible in as limited hardware.
You're sort of missing my point - the goal here is to cram the most optimized models - that is, the models that are going to best allow them to do meshed swarm coordination, or whatever. Larger is not always better; better is better.
I have no particular reason to think that the direction LLMs research is currently developing is good at all for that sort of thing. Cramming Astra into a drone is going to result in something that's much slower and less decisive than a simple deterministic script, because LLMs are compute-hungry. That's not to say that neural networks won't be helpful - computer vision is great, for instance.
But LLMs are not the only form of artificial intelligence, and for most military tasks a strong general-purpose LLM is both overengineered and poorly suited for the mission, especially the mission of terminal target determination. For intelligence synthesis, they likely have some promise.
Shrike
1d ago
All the world militaries and terrorist groups have great interest in making full fledged AI work in the limited hardware a drone possesses.
Eh, in the sense that the AI can do independent target discrimination, sure, but that's already a solved problem (LRASM can do this, for instance) although it can be iterated out further.
World militaries and terrorist groups have no interest in making sure their drone processor can write poetry, or whatever.
Shrike
1d ago
A conventional war in an environment saturated with large underwater drones, perhaps.
Perhaps (particularly since the US is looking into a low-cost heavyweight) but...over what?
I maintain that there are no plans to have a nuclear war or any war over Taiwan.
This is a sort of ambiguous statement. Are you saying that the US does not intend to start a war over Taiwan, or that it has no contingencies in place for waging a war over Taiwan?
Shrike
1d ago
Nobody is getting ready to stop the Chinese invasion.
You can tell what war a country is planning to fight by following the money to their munitions procurement. In the case of the United States, if you follow the money, you will notice (among other things) that they intend to increase production of torpedoes.
Are they likely to be low on torpedo stockpiles? No; a quick look at Wikipedia tells you that they had 1,000 heavyweight torpedoes stockpiled in 2001, they have continued production at a low rate since then (50/year as of 2017, Wikipedia thinks, so you can estimate total inventories somewhere between 1,000 and 2,000).
Background knowledge tells you that training torpedoes are reusable, so their burn rate from training is low. The entire US submarine fleet can only hold 1,500 torpedoes or so, even if they were all deployed at once (they won't be; a quick Google tells you that 20ish attack submarines will be out of service at any one time) and were only carrying torpedoes (they won't be).
The entire Russian Navy is about 70 submarines and 100 major fleet combatants. The entire Chinese Navy is about 75 submarines and fewer than 200 major surface vessels. (Not including minesweepers or amphibious assault/transport ships in the count here, or, say, patrol boats, but including some low-capability combatants such as corvettes.) In any conflict involving one or both of those opponents, the US would not rely on its submarines alone to sink their ships, and it would be fighting alongside allies.
So one can conclude that the US is preparing to fight a war against ocean-going opponent(s) at a scale so grand that they anticipate their attack submarines will expend their entire onboard munitions stockpile, return to port, reload, and sally forth with more torpedoes.
Based on that evidence, what exactly would you say they are "getting ready" for?
Shrike
2d ago
Yeah, as a religious and relatively conservative person, I would say that the natural role of a parent is to take care of their child and a child to be instructed by their parent. Thus, while the parent (or rarely the child) can do something bad enough to warrant severing that bond, severing the parental bond is itself an act of violence and should only be done in extreme circumstances.
Thus I wouldn't want to take kids from "woke communist parents" or whatever - even if I was absolutely convinced that they were wrong, taking kids away from their parents is a good way to convince them you are wrong. Whereas I think that a good parent-child relationship and familial relationship serves as a model and a stepping-off point for other relations in society. Turning someone towards religious truth (or social or political truth, to the extent that there is such a thing) is, I think, more difficult if their first relationship (with their parents) is disordered. Thus, while a sufficiently disordered relationship may be cause for separation, disordering a sufficiently ordered relationship may itself make a person's relationship with society (and God) more difficult.
Hopefully that makes some sense.
Shrike
3d ago
Also he left OpenAI well before they expected to be making any money at all, all the way back in 2021.
So he left his job as an employee at a nonprofit to start a for-profit organization and I am not supposed to believe this is an obvious financial trade-up?
I don't think humans are entirely motivated by money. But (at least in America) quitting your job to make more money is quite routine.
Which narrative is that?
I'm not sure this means chooky is wrong – training and inference are different tasks, and so don't (necessarily) run on the same datacenters/GPUs. So chooky could be correct that they are slowing down on unsustainable capex (training) but that doesn't necessarily mean that pausing unavoidably means less capex spend total, because you could pivot to building inference datacenters to serve customers with existing models, shifting from expensive R&D to increasing revenue via selling inference. Right?
Now, caveat – I don't know of the top of my head what the current datacenter build out looks like – how much of it is inference versus training, what's being financed directly by the primes versus third parties, etc. etc. I gather that you're not familiar with it either (and I'm not sure it's all public knowledge, anyway). So I am just thinking about the logical levers at play here.
Shrike
3d ago
pausing unavoidably causes less capex spend
Why would this be the case, in your telling? Why wouldn't AI companies pivot towards customer compute build-out?
Shrike
3d ago
Is Amodei making more or less money now than when he resigned from OpenAI complaining about safety?
I think the quietness of modern submarines makes the GIUK gap more important. With a quiet submarine in, say, the middle of the Pacific or the Atlantic, it could be anywhere and finding it could be tricky. The GIUK gap is (from what I understand) a bit more shallow than the middle of the ocean and it is a much more finite space. If you can carefully observe it you can pick up submarines and try to track them before it becomes a needle-in-a-haystack race.
If there's anything that makes GIUK less important it would be US technological developments making it easier to find submarines in the middle of nowhere. And some of the Cold War SOSUS intelligence suggest US capabilities can be scary in this regard. But I am not sure how reliable that is (there are a lot of technologies that are amazing in ideal conditions and quickly turn to mash if you look at them the wrong way) or how much more recent quieting measures have eaten into it.
More options
Context Copy link