Caleb is a mad genius (or Yet Another Fan Club Thread)

[Jace911]

Just got around to listening to Erasure Squad, and I have to say my favorite bit was easily the death of Igor.

"I got i-" "I got-" "I got it-"
*DEAFENING AUTOMATIC GUNFIRE*

Although "prove you're a real human" was also pretty hilarious. I'll have to borrow that for future EP games. :p

[Claive]

I cannot wait to listen to Caleb run Base Raiders!  It will be glorious!

[RadioactiveBeer]

Just got around to listening to Erasure Squad, and I have to say my favorite bit was easily the death of Igor.

Although "prove you're a real human" was also pretty hilarious. I'll have to borrow that for future EP games. :p

Yeah, a kind of inverse-Turing test. It was definitely brain-fodder, along with Augustine and "how do we make a good/not-actively-genocidal seed AGI?" ideas for something in my current campaign, where they try to teach a seed AGI empathy through a kind of modified Turing test.

[Jace911]

Yeah, a kind of inverse-Turing test. It was definitely brain-fodder, along with Augustine and "how do we make a good/not-actively-genocidal seed AGI?" ideas for something in my current campaign, where they try to teach a seed AGI empathy through a kind of modified Turing test.

If you're looking for more ideas along those lines I'd suggest My Invisible Friend by Anders Sandberg; it's about scientists trying to teach a Seed AI human morality by slaving it to actual living children and forcing it to watch them mature and develop socially and mentally.

[RadioactiveBeer]

I'll definitely check it out. My current concept is prospectively titled "Trinity" because names are hard and it's made of three parts so whatever.

Parts 1 and 2 are the seed AI's, carefully constructed and slightly hobbled in their hardware so that they lack the capacity for improving themselves but have the capacity to improve the other. So 1 cannot improve 1, it can only improve 2 and the vice versa. Currently, the AIs are named Romulus and Remus.

The third component is a human overseer. The overseer bridges the gap between Romulus and Remus, permitting or preventing the improvements from going through. The human overseer can communicate with the AIs and will ask them to justify why each change should be permitted and experiment parameters are to 'reward' selfless acts through narcoalgorithms.

The idea is that the AIs learn empathy and to value others by having their improvements come from another party, encourage them to help others and 'train' them via neural network learning patterns to think in terms of morality, helping others etc. Of course, a nice bonus to that would also be that they'd learn to see humans as authority figures and at the end of it you have not one but two god-machines on your side. The downside is of course that it's slower due to the human factor.

Currently, I'm thinking of ways it can go wrong (because otherwise whoops no story). My main idea of how it goes wrong is this: while neural networking allows AIs to learn 'organically' in a way similar to people, the problem is that no matter what data you feed them and how you train them, they aren't necessarily learning the lesson you think you're teaching them. The classic example for this is a real-world satellite AI I read about that was fed different pictures, some of which had concealed tanks in, to train it to spot hidden enemy vehicles. However, what it actually 'learned' was the ability to differentiate the pictures based on light levels in the pictures because the darker pictures were taken after the tanks had been put in place.

So essentially, a failing of the Trinity set-up is that rather than teaching the AIs to value humans, it teaches them how to manipulate humans and set up very Machiavellian mindsets - helping others only in ways that allow you to benefit.

[Jace911]

I was going to suggest that the lesson the AIs learn is "improving people is nice, so that's what we'll do!" without actually explicitly defining just what "improvement" means, so when the Firewall team inevitably shows up to find out how the experiment fucked up they find a station full of people who have been "improved" in various horrific ways: cyberbrains spliced together to form a gestalt consciousness that can think faster and analyze problems from multiple angles, xenomorph specimens abound, fresh Watts-Macleod carriers who have gone insane from having their minds ripped open, etc etc. That might be too Dead Space depending on the tone of your game, though.

[RadioactiveBeer]

That is one option, though I don't want to come across as too one-note and the last mission the players had involved a fun trek in the ruins under New Mumbai where stuff was, you know, awful in a Dead Space-y kind of way. (Thanks, Zone Stalkers 'random fucked-up-shit table'!) So it is possible that taking it that route would be a bit too much Dead Space at one time. I can spin it out a bit with some social and investigation stuff to GET to Trinity and I'm honestly a bit tempted to throw the whole thing out to replace it with the Oracle AI from Think Before Asking - a great example of a scary IDEA.

The other option they might take (though given their mentalities I think they recognize the AI as more urgent) involves Pax Familia infiltrating LLA banks to cash out massive amounts of rep and snap two economies over her knee simultaneously.

[Jace911]

If you want to keep it from getting too Dead Space-y, instead of having space zombies for them to fight you could just have the station be full of people who have been "improved" but don't actually attack them; the real threat comes from the Seed AIs trying to corral them into the medbay with the station's systems so that it can "improve" them as well. For additional scary idea points, make the improved people out to be obviously inhuman and horrific to the sentinels while also emphasizing that, technically, they are better than most transhumans and actually seem happy with their current state in a twisted sort of way. Be sure to underline the fact that, when you get right down to it, what the Seed AIs are doing is the logical extrapolation of the entire transhuman ideal; it's the same principle as loading yourself up with biomods or nanoware, just multiplied by an order of magnitude or two.

[Claive]

"To the Room!" new favorite quote.

[clockworkjoe]

"To the Room!" new favorite quote.

what is that from?

[Claive]

"To the Room!" new favorite quote.

what is that from?

It is from the Call of Cthulhu: The Sense of the Sleight of Hand Man episode 4, When they are going to go do drugs.

Edit: 2:04: 10 ish

[HooliganTuesday]

I couldn't think of anywhere better to put this but I thought that Caleb and Ross would love it and wanted to bring it to their attention.

It's a six part British T.V series called Utopia, it's an awesome series about a group of misfits who accidentally end up on the wrong side of a ruthless conspiracy. It's hard to say more about it without giving the plot away, but the show feels like Caleb decided to sit down and write a Nights Black Agents game and leave out the vampires. The youtube link is the opening scene of the first episode and it does a much better job of selling the show than i'm failing to do here.

https://www.youtube.com/watch?v=01VGtX9xfh0

[Vivax]

It's probably indicative of Caleb's mad ingenuousness that his speech patterns have infected me while I GM. Then again I've been exclusively GMing Caleb's scenarios to alternatively horrified/thrilled players. That said, if Caleb's games are some kind of weird basilisk hack that slowly turns me into a larval Caleb I would not be surprised.

[Kamen]

Have you been giving your players options by starting sentences with "So here's the thing..."? Speaking of actually, whenever my group gets around to playing Numenera I have ideas for cyphers, the magic items in-world, based around Caleb and Ross: The Stokesian Trap Detector and the Rossian Choice Analyzer! When I run upon a potential deathtrap(or traps) the astral voice of Mr. Stokes rings in my ears, "So here's the thing..." and tells me about the dangers ahead, but only like once a day or so. The thing is, no matter what, you're kinda boned.

[Vivax]

It's mostly that I keep using the word "horrific" and use emphasis the same way he does. There's also "You can do it, but..." which interjects some PC decisions when they've forgotten something that their characters would have remembered.