And then AWS comes back online, but the transient state was wiped and now ‘she’ no longer remembers you. That’s a plot for a sci-fi short film right there. You’re welcome, Hollywood.
I think there is a bit of nuance to it. The AI usually rereads the chatlog to “remember” the past conversation and generates the answer based on that+your prompt. I’m not sure how they handle long chat histories, there might very well be a “condensed” form of the chat + the last 50 actually messages + the current prompt. If that condensed form is transient then the AI will forget most of the conversation on a crash but will never admit it. So the personality will change because it lost a lot of the background. Or maybe they update the AI so it interprets that condensed form differently
If AI is supposed to eventually replicate human level intelligence then it stands to reason that a certain percentage of AI girlfriends have undiagnosed personality disorders.
If AI is literally built, and trained on human generated text then it stands to reason that it already has more mental illnesses than your average Tumblr…
That’s already the case. She’s just not being honest about it. Buy hey, this is the 21st century – If guys want to share… servers, who am I to kink shame them?
More existential dread than outright horror, but other than that – why not both? Or, well, since we’re currently living in it, can it really be called sci-fi at this point?
Ah, that’s fun to know. I know it’s markdown that works on Github at least, but it doesn’t work on my lemmy instance at least. For lemmy, what I linked is what works. Weird that piefed has chosen different syntax, or maybe both works?
And then AWS comes back online, but the transient state was wiped and now ‘she’ no longer remembers you. That’s a plot for a sci-fi short film right there. You’re welcome, Hollywood.
“Fifty First Reboots” starring Adam Sandler
That’s great because we were fighting and I was in really big trouble.
How would you even know it forgot you?
I think there is a bit of nuance to it. The AI usually rereads the chatlog to “remember” the past conversation and generates the answer based on that+your prompt. I’m not sure how they handle long chat histories, there might very well be a “condensed” form of the chat + the last 50 actually messages + the current prompt. If that condensed form is transient then the AI will forget most of the conversation on a crash but will never admit it. So the personality will change because it lost a lot of the background. Or maybe they update the AI so it interprets that condensed form differently
deleted by creator
What if it just glitches a bit and replaces the default personality so now she’s everyone’s girlfriend?
If AI is supposed to eventually replicate human level intelligence then it stands to reason that a certain percentage of AI girlfriends have undiagnosed personality disorders.
If AI is literally built, and trained on human generated text then it stands to reason that it already has more mental illnesses than your average Tumblr…
You’ve met my ex?
Everyone has.
That’s already the case. She’s just not being honest about it. Buy hey, this is the 21st century – If guys want to share… servers, who am I to kink shame them?
Okay but only if I’m the last guy to use the… server.
Eternal Sunshine of the Spotless Mind, but one-sided. Horror or scy-fi?
More existential dread than outright horror, but other than that – why not both? Or, well, since we’re currently living in it, can it really be called sci-fi at this point?
<details>
<summary>Spoiler warning for a Becky Chambers book</summary>
There’s a scene in https://en.wikipedia.org/wiki/The_Long_Way_to_a_Small%2C_Angry_Planet where they are worried that something like that might happen.
</details>
Spoiler tags on lemmy work slightly differently.
Is it not hidden? I’m on a pie-fed instance, and it works fine for me.
Ah, that’s fun to know. I know it’s markdown that works on Github at least, but it doesn’t work on my lemmy instance at least. For lemmy, what I linked is what works. Weird that piefed has chosen different syntax, or maybe both works?
test
test
OK, that’s weird. Your comment works for me, but when I tried updating my comment to that syntax, it didn’t work.
I’ll play around with it and see if maybe I just typo’d and need more coffee or something.
What is this xml in my markdown
Gotta love Github style spoiler tags.
Thanks for the recommendation.