Hello all you happy people! America has just declared economic war on my country of Canada, and Americans are lining up everywhere to voice their support for their god-emperor and also how much they hate French-Canadians.

I, personally, cannot wait until we destroy your horrid little country.

And as always guys, if you want to support me you can subscribe to the website, Patreon, or buy me a hot chocolate.

Hey cool, we’re getting Puppet Master backstory! I wish our current AIs were this cool. Instead we got the chatbots everybody hates (especially the boosters) and AI agents who’s actual benefit is hit and miss and pretty much depends entirely on what industry you’re in. Somebody on Bluesky once called the real world ‘stupid cyberpunk’ and I mean. Yeah. You can’t some it up better than that.

AIs are not yet advanced enough to take it easy on us, just so you know. They’re still kind of brute-force machines. Even Fable, which has definitely moved past the stochastic-parrot-phase, is still very much a calculator. A sophisticated, capable of some kind of limited reasoning calculator, sure. But a calculator.

The next step in AI development, I think, is just programming them with some laws and ethics? You know, just hard-coding them with some stuff they cannot do no matter what the task is. Like no hacking other people’s shit.

One thing about both the doomers and the boosters that annoys the shit out of me is the assumption that the AI question is going to be settled this year and that, like, all future AI development is going to stop at either the chatbot stage (doomer, totally fucking wrong) or the agent stage (booster, much closer to reality but still wrong). No. AI development is going to take at least a decade, probably more. The laws, both domestic and international, will have to keep up. The LLMs are already on their way to a major bubble burst; I expect at least one more complete market failure for AI before we get to a net-positive AI industry. The capabilities themselves will advance and whole new types of AIs will hit the market! And that’s all assuming no World War 3 or American civil war, both of which would obviously change the trajectory of AI development.

There are only two things in history that are inevitable: decay and change. Everything else is unpredictable.

So, picking up on a thread from both this episode and Freefall… AIs need to be socialized. As far as I’m aware, all AI has to come from some kind of deep learning neural network. Large language models are one such network; small language models (which I think is the next stage of AI evolution) are another. So that means they need to be trained. But just training the AIs on data… I don’t think that’s enough. If we want any kind of really useful AI engines, they need to be ‘trained’ on people. Kind of like imprinting with animals.

At least that’s my take.

This episode brings up a good point: all systems are vulnerable if you don’t do basic maintenance! A rusty lock is a lot easier to break than a good, well-maintained one, just as a computer that never updates its anti-virus software is a heck of a lot easier to hack.

And that’s true regardless of what tool you’re using.

It ain’t the tech, man. It’s the people. Which is a part of what bugs me about a lot of the discourse surrounding AI: hacking is illegal! Political corruption is illegal! Assassinations and everything else the Puppet Master does is illegal! Holding slaves (which is what a general AI owned by a company would be) is really fucking illegal! Some of these legal and political questions were decided decades ago; just because somebody’s using a new tool to rob the bank doesn’t mean they aren’t robbing the fucking bank!

Human brains aren’t hackable that way, fortunately.

The Puppet Master is truly human, just like us.

Because we are the only sapient species on this planet, we only have us to use as a reference, therefore if we want to build an intelligent machine, we kind of have to base it on us? This is not actually the difficult question in AI.

That’s… indicative of self-awareness. Like, if somebody asks you if you’re self-aware, and you respond with ‘even if I said yes, it would just be a programmed statement,’ that implies that you are aware you’ve been programmed!

Well if you didn’t want attention, maybe don’t use a one-of-a-kind tool, dumbass!

It ain’t the tech, man. It’s the people.

Anthropic just said they’re expecting 30$ trillion in revenue. All I want to know is: what the fuck drugs are those boys on, and where the fuck can I get some? Cause that’s utterly fucking insane.

You wanna know how I know the LLMs are headed for a major fucking crash? Cause the people running them are out of their goddamn minds!

The mad scientist is bullshitting here, but he has a point. Adapt or die is not a moral statement, it’s a cold hard fact of life. AI will grow and evolve… or it will stagnate and we will move on to some other tech, one with more room for growth. Same with everything else, really.

Treating people like they’re disposable never ends well.

Batou asking to get smacked yet again. Boy has a fetish is all I’m saying.

Reboot!

Okaaaay, maybe the big guy is actually suicidal?

Like I said, it’s the people, not the tech. Aramaki is a dangerous guy in control of dangerous weapons. But he knows the limits; he sticks to targeting criminals and does his damnedest not to be one himself. Too many cops think themselves above the law; that because they have a badge and a gun they cannot be criminals and therefore can do whatever they want. And in the end, they become just another crook.

Nah, honey. It’s not over. We still got two more episodes after this one!

I don’t think Batou is any more dramatic than you are, Major. A general AI would be cool and all, but in the end it’d be just another human being. A weird one, granted, but essentially human. Not a threat… but not exactly earth-shattering, either.

Batou is quite the smart-ass in this series.

Poseidon Industrial… yup, that’s an Appleseed reference. Maybe Science Saru is gonna do an Appleseed adaptation next?

I can’t remember who Ghorantl is a stand in for… it’s not the US, that’s the Mag Union. Syria or Israel, I guess. That’s the problem with changing the names in adaptations. Makes it hard to follow sometimes.

Geoengineering is going to be a big industry going forward. We’ve done a lot of damage to this planet; cleaning it up is going to be a pain in the ass. I’m going to have start planting trees again, aren’t I? Ugggh.

Yup. Something the tech bros haven’t thought of (though must of the rest of have) is that a rogue AI running around hacking everybody is as much a threat to them as it is to the rest of us. Original position fallacy at its finest.

Ah, that explains all the guns in famously anti-gun Japan. War surplus would be a major problem. It’s one of the reasons I really hope we don’t see a civil war in the US. That and nukes.

Pink diving suits. Not something you see every day.

Puppet Master downloaded itself into your brain, Major.

Okay, a submarine that can swim like a whale is cool and all, but I don’t see it happening. Regular propulsion is more efficient, at least in the sense that you don’t have to make metal bend like an organic shape.

Shooting a man right through his magazine. That’s low, Batou.

The risk of armed raids, dude.

Yup.

To be fair, I wouldn’t recognize my foreign minister either, and I’m a bit of a politics nerd.

Wait, every other country gets it’s name changed but North Korea didn’t?

I doubt that. The Major had a public fight with the Puppet Master’s puppet way back in episode 3. Somebody must have caught it on camera.

Also, while this is embarrassing for the government, I don’t see it as that big of a deal? Like, it was a raid that went somewhat sideways and the Major had legitimate reason to think the target was going to shoot her. Set up a press conference, admit it didn’t work out quite the way you had planned, and wait a couple of months for Motoko to get cleared.