Fishin' in the stream of consciousness (all-purpose, no topic chat thread)

Theres a some rat folk here on kintsugi, this is for you

 

What precious friends, thank you for sharing!! <3

 

I found a Kintsugi-esque survivor or mental health support group makes for a really good setting for a story for the same reason a bar does; as someone once put it, you have a ready-made reason why a diverse cast of characters would show up there.

 

I cant believe its Neil Day again already

Godspeed little naked rat, bangin out those tunes

 

This looks like fun. I've been wanting to do a Rimworld generation ship.

 

listened to the whole talk, its interesting

 

I found the classification of goals thing to be personally helpful. I have a little bit of money, and I'm trying to decide my unit build order for real life. Do I get the motorcycle that I want, just because I want it and it would make me happy, or do I save up and start a business because it could make me happy later? The realization that getting the motorcycle is a valid goal just on it's own is helpful.

 

That potbellied guy in the video thumnail image is actually a far more accurate depiction of me.

 

I think it's more like humans and other animals are programmed to want to be happy, and everything else we can think of that we think might make us happy is an intermediate goal. If the AI is hard-coded such that stamp count == level of happiness, then it will not reevaluate that goal any more than a human will question whether or not it wants to be happy. We're hard-coded to strive for the release of certain chemicals in our brains that tell us we're happy. We do frequently reevaluate our intermediate goals, as in what we think will make us happy, but being happy is always the ultimate goal. This is why anything that directly edits the happiness counter (like drugs) is dangerous.

 

Perhaps, depending on how reliably you can program a general AI. Leaving the absurdity of deploying a superintelligent AI to collect stamps, 'Collect Stamps' might be its overarching goal, or it might be closer to reproduction for biological creatures. Or maybe 'Stamp' is just at the top of its Maslow hierarchy.

Well, that's a valid question in itself. Several, actually. Though even if it doesn't have self-awareness, it might change if it's capable of analysis.

And the AI, being superintelligent, would likely realize it can literally do this. It could alter its own code to lock the happiness counter to maximum, or change it to be based on whatever, or...

It's also worth noting that people do occasionally decide other things are more important than their own happiness. So given an even slightly anthropomorphic entity, it's far from unthinkable that it might decide to do something 'aversive' for a variety of reasons.