It's an interesting situation in Ukraine.

Ukraine has essentially run out of artillery shells and anti-missiles. That's not an absolute measure, but effectively Russia is missile striking all the infrastructure in Ukraine, notably power generation facilities, with impunity.

The Russians are also slowly taking territory. You might have heard about the recent fall of Avdiivka, which is officially a win for Russia except that Ukraine made it a very expensive piece of real estate. I've heard one estimate that Russian casualties are 10:1 against Ukrainian, so it's really a win for Ukraine. Except that Russia has so many people it can throw into the war effort it might not make a difference.

On the flip side, Ukraine has damaged several oil processing facilities inside Russia 200 miles East of Moscow. Two soldiers carrying small drones in a backpack can hike across the border, deliver a small munition (probably more than a hand grenade but not much more) right to the vertical distillation column using video feedback for targeting, and the distillation column is an integral part of the process and the most difficult piece to repair.

Ukraine has taken some 14% of Russian oil processing offline using this method, which is a huge bite out of Russia's federal budget. Also, Russia now has to allocate resources to protecting vital infrastructure all over Russia.

Ukraine has also had good luck with water-based drones: put a bunch of munitions on a motorboat with a GPS and video feedback for targeting, paint it black and send it at night, several hundred miles with pinpoint precision to sink a warship. Russia discovered experimentally that all of their anti-whatever guns are intended for incoming missiles and other ships, and so they can't point down low enough to hit a small motorboat within striking range. You have to get the crew to shoot at the drone from the deck with rifles and hope you hit something important.

Ukraine has basically kicked the black sea fleet out of the western half of the black sea using this method.

Of note, these drones are being built in Ukraine by Ukrainians. They're not donations from other governments.

Ukraine now has lots and lots of military observers from various countries across the world looking in on the military aspects of drone warfare, which is a completely new tactic for war. If it takes an anti-missile costing $100,000 to take out a drone costing $1,000, that's an obvious advantage to the side using drones.

And no one has tried drone swarms yet either, and I think that would be the next logical step. Exhaust your opponent's anti-missile shield over the city with one wave of cheap drones, then send in the second wave with incendiary munitions to set everything on fire all at once.

And all for the price of 1 anti-missile missile.

In the case of Ukraine, the success rate is very high because anybody in range is likely an enemy soldier.

Israel's success rate may be as low as 0.1%. That tells us that robots can't tell civilians from military. A large enough stockpile of human shields would be a serious problem.

And we know drones et al are vulnerable to GPS spoof attacks, making such an attack risky against a technologically advanced enemy with intellectuals and engineers forming a scientific take on special forces.

Why bother with a missile? You're here, so a geek. You know GPS jamming is effective, as is GPS spoofing. All you need is a parabolic dish and a high power transmitter. There's simply no possibility of a wide-angle transmitter on a satellite matching a narrow beam that's broadcast from a hundredth of the distance. Sure, there'll be authentication keys. And social engineers have compromised most of the world's governments, which means the keys will be for sale somewhere.

The only way I can the robot army being effective is if they flatten everything at long range, indiscriminately. And that is going to cause its own problems. Especially if the software gets hacked prior to install. Which will happen, because hiring and training an army of hackers in Mitnick-style social engineering tactics costs a tiny, tiny fraction of the expense of maintaining a wall of tactical nukes that can EMP the robot forces.

The robots work OK, but the AI doesn't. Israel is using AI extensively to target Hamas at the moment, with the very best AI that exists and the very best military minds the world can produce. The success rate is somewhere between 1% and 0.1%.

Face scanning tech also depends on the data set being valid. The DOD has been compromised many times by airwall violations, security violations, improper screening, and extremely buggy software from Cisco and Microsoft.

All the enemy needs to do is write a rootkit that flips a couple of bits. The robot army now faces the other way and friends are identified as foe. I wouldn't put it past a group like the Lazarus hackers to be capable of such a stunt. We already know the enemy is capable of GPS jamming and GPS spoofing, because they've done so to hijack US drones, and that's another potential vulnerability.

US military robots are also known to have severe problems identifying that a person dressed as a tree is a person, not a tree. A skillful enemy could walk through US robot army lines without impediment, unless the US robots shoot indiscriminately. But if the US robots are genocidal, mutually assured destruction becomes a viable tactic. You can't be more than dead, after all.

And if the US includes a death switch, given that US defence contractors don't always wipe hard drives and the military don't psychologically screen very well (Manning was known to be seriously mentally unstable prior to deployment, for example), there's absolutely no guarantee the enemy won't simply learn it and spoof it.

I just don't see how the US think this could possibly work.

Its success rate in Israel stands at somewhere between 1% and 0.1%.

One gun can shoot at one target at any one time. If your AI-guided robot army is shooting up chicken farmers and goat herders, it's ergo not shooting at the army that's flanked it which threatens to overrun the opposing side's now largely undefended turf.

A robot army can also be taken out by EMP weapons - basically tax nukes. Since robots can't distinguish between soldiers, civilians, and cake stands (AI is pretty dumb), the defending side already faces complete genocide. You can't get any deader than that, so there's no incentive to not flatten the enemy with nukes and a very slim chance they won't fire back, because it's hard to maintain an expensive nuclear defence and an extremely expensive robot army at the same time.

(Basically, same reason the US is now outgunned on fighters, the new ones are so expensive they can't afford that many. The US relies utterly on them being more destructive faster, but again, what's the point in NOT invoking MAD when your enemy has demonstrated they're genocidal and no respectors of the norms and laws of war?)

Your legal theory where a person can break into someone's house, take book that the person doesn't have the right to share, photocopy it, and share it, and that this is the VICTIM's liability because their locks aren't good enough to stop the attacker, is duly noted.

Or is this just a good old fashioned human-dupe?

You want a pre-WW2 suburb.

I was visiting Oxford UK on business and I stayed at a colleague's house which dated from the1800s. I was shocked that the front door of her house was right at the sidewalk, you could look right into her front room. But it turned out that by giving up privacy in that front room, she got an enormous and very private back yard. The arrangement was something like this. That's just a street in the area I randomly picked off of Google Maps satellite view, but I checked it for walkability: it's less than one minute's walk from the local boozer, and on the way back you can get a takeaway curry.

I'll quote from the Wikipedia Article: "In urban planning, walkability is the accessibility of amenities by foot." It is important to contrast this with the practices it was intended to counter (again from the same article): "... urban spaces should be more than just transport corridors designed for maximum vehicle throughput."

Transit is an integral part of walkable planning simply because it gets people *into* neighborhoods so they can do things on foot. But cars are a way to get people into an area too, so cars can and should be part of *walkability* planning. For example there's a main street area near me with maybe 50-70 stores. When I visit I contribute to congestion by driving around looking for a parking spot. A carefully placed parking lot could reduce car congestion on the street while increasing foot traffic and boosting both business and town tax revenues.

This is in turn also not correct. All works are NOT automatically granted copyright. The work has to meet certain qualifying standards, for example more than de minimis human creative work. You can't just write "My dog farted" and assert that it's copyrighted; that simply won't pass creativity standards. Some works, such as AI works (which have not been not further human processed or involved in a creative selection process), are automatically denied copyright on these grounds. A wide variety of things are also not available for copyright protection - ideas, facts, short phrases / slogans, government works (with certain exceptions), and so forth are not copyrighted. Also, works posted online - aka, virtually all works anyone in this discussion is talking about - are generally posted on sites with a TOS, which requires the user granting the site at least limited distribution rights (and in some cases, full rights over the work).

And it's BTW a good thing that de minimis works are ruled out, because so much of our online life is basically structured around copyvio. For example, the "Forward" button on an email client might as well be labeled "Violate Copyright" - you're taking someone else's work and sending it to a third party, generally without the author's express consent. The primary defense that one has in this case is to argue that the received email e.g. lacks sufficient creativity, is just facts and ideas, or so forth.

VMS has a versioning filesystems, which is a useful thing to have. It's also pretty secure (once you disable scripts in mail subject headers).

I'd contend it has a role in server systems, but really it's a struggle to think where else it would fit.

Running your own website may get you past a TOS, but it doesn't mean you can disclaim fair use.

LLM training falls outside many of the tests commonly applied to decide fair use.

If Google can win Authors Guild, Inc. v. Google, Inc., there is no way AI training would run afoul.

Google: Ignored the explicit written request of the rightholders
AI training: generally honours opt out requests

Google: Incorporated exact copies of all the data into their product
AI training: only data seen commonly repeated generally gets memorized, otherwise it just learns interrelationships

Google: Zero barriers to looking up exact copies of whole paragraphs or even whole pages of the copyrighted works.
AI training: Extensive barriers set up during the finetune; success at extracting said information has required attack vectors, frequently estoteric, and sometimes requiring the attacker to provide part of the copyrighted text themselves.

Google: Product literally designed for one purpose, that purpose being to return exact content
AI training: Literally the opposite; designed for *synthesis*, for solving *novel* tasks. .. and ***Google won***. Google Books was found to be a "transformative use". There is NO way that Google Books is "transformative" but LLMs are not.

Or take diffusion models. The amount of data on the weights is on the order of one byte per training image (give or take an order of magnitude). Meanwhile, Google Images searches return 50 kilopixel scaled copies of *exact copyrighted images*

The simple fact is that the very existence of the internet relies on the fact that automated processing of copyrighted data to create new transformative products and services is fair use.

Back at you.

I'll repeat: they were creating attacks against the models to find ways - often quite convoluted and esoteric - to trick them into doing things that their designers put safeguards against them doing, as you can quite literally see in my examples above.

People who write this sort of stuff remind me so much of the people who share viral messages on Facebook stating that Facebook doesn't have the right to their data, and that by posting some notice with the right legalese words they can ban Facebook for using their data. Sorry, but you gave up that right when you agreed to use their service, and no magic words are just going to give it to you.

(Let alone when talking about rights that you never had in the first place, such as to restrict fair use)