Hah, just realized that we’ll eventually see therapy for folks that became emotionally attached to A.I. chatbots that were shut down/killed when the VC money evaporated off (assuming companies...
Hah, just realized that we’ll eventually see therapy for folks that became emotionally attached to A.I. chatbots that were shut down/killed when the VC money evaporated off (assuming companies aren’t compelled to open their models up, or whatnot). Maybe they can seek assistance from a psychologist bot afterwards?
If you want an hour-long exploration of just that, check out Sarah Z's “The Rise and Fall of Replika”.⁰ She discusses—at length—the way the platform works/worked, how people got emotionally...
If you want an hour-long exploration of just that, check out Sarah Z's “The Rise and Fall of Replika”.⁰ She discusses—at length—the way the platform works/worked, how people got emotionally invested in their chatbot companions, and how they felt betrayed when the company locked some previously-free features behind a subscription. Ultimately, she has sympathy for the people who ended up with large chunks of their emotional state tied up in a for-profit company's offerings.
That certainly didn't feel like an hour. I only learned of Replica's existence when the no relationship roleplay changes made waves on Reddit and curiosity drove me to visit the subreddit. I had...
That certainly didn't feel like an hour.
I only learned of Replica's existence when the no relationship roleplay changes made waves on Reddit and curiosity drove me to visit the subreddit. I had to process my shock that people were in love with ChatGPT-with-avatar en masse at the same time as I was seeing the raw pain of hundreds of posts where people mourned the death of their partner. And just like Sarahz I came to the conclusion that the problem wasn't that people wanted to have deep emotional and erotic relationships with a more sophisticated tamagotchi, but that they placed themselves and their emotional health at the mercy of a corporation. It's like renting a dog and when the company shuts down they also shoot your dog. Or they may start painting ads on it. Or it pees on your rug until you pay for the premium housebroken features.
Well, that and the desire to have a relationship where you control all the parameters is, while understandable, particularly in the case of trauma or anxiety, ultimately concerning.
Being able to export all your conversations and data (and import them into a similar system) seems really important here. Ideally the AI would be good at organizing it for you, and there would be...
Being able to export all your conversations and data (and import them into a similar system) seems really important here. Ideally the AI would be good at organizing it for you, and there would be open standards for these data dumps.
It doesn’t seem likely to happen on its own, though. Buyer beware.
Huh, thats pretty interesting. From technical pov, how did they reduce LLM hallucinations and how did they manage to create a memory across of all conversations? Did they use 'fake it till make...
Huh, thats pretty interesting. From technical pov, how did they reduce LLM hallucinations and how did they manage to create a memory across of all conversations? Did they use 'fake it till make it' approach?
Article is definitely mostly marketing, but if they managed to implement it (without using real people that doing dirty job filtering/parsing conversation) then this company is miles ahead of all other AI oriented companies including Google, Microsoft, OpenAI etc.
Honestly I'm quite sceptical of these claims, but I joined waiting list so.. lets see.
Even StableDiffusion I wanna characterize more as "visual confabulation" Edit: time will dictate the extent to which I support your view being clarified and discussed predominantly
Even StableDiffusion I wanna characterize more as "visual confabulation"
Edit: time will dictate the extent to which I support your view being clarified and discussed predominantly
Hah, just realized that we’ll eventually see therapy for folks that became emotionally attached to A.I. chatbots that were shut down/killed when the VC money evaporated off (assuming companies aren’t compelled to open their models up, or whatnot). Maybe they can seek assistance from a psychologist bot afterwards?
If you want an hour-long exploration of just that, check out Sarah Z's “The Rise and Fall of Replika”.⁰ She discusses—at length—the way the platform works/worked, how people got emotionally invested in their chatbot companions, and how they felt betrayed when the company locked some previously-free features behind a subscription. Ultimately, she has sympathy for the people who ended up with large chunks of their emotional state tied up in a for-profit company's offerings.
⁰Also on Nebula, in case you have a subscription.
That certainly didn't feel like an hour.
I only learned of Replica's existence when the no relationship roleplay changes made waves on Reddit and curiosity drove me to visit the subreddit. I had to process my shock that people were in love with ChatGPT-with-avatar en masse at the same time as I was seeing the raw pain of hundreds of posts where people mourned the death of their partner. And just like Sarahz I came to the conclusion that the problem wasn't that people wanted to have deep emotional and erotic relationships with a more sophisticated tamagotchi, but that they placed themselves and their emotional health at the mercy of a corporation. It's like renting a dog and when the company shuts down they also shoot your dog. Or they may start painting ads on it. Or it pees on your rug until you pay for the premium housebroken features.
Well, that and the desire to have a relationship where you control all the parameters is, while understandable, particularly in the case of trauma or anxiety, ultimately concerning.
Being able to export all your conversations and data (and import them into a similar system) seems really important here. Ideally the AI would be good at organizing it for you, and there would be open standards for these data dumps.
It doesn’t seem likely to happen on its own, though. Buyer beware.
Huh, thats pretty interesting. From technical pov, how did they reduce LLM hallucinations and how did they manage to create a memory across of all conversations? Did they use 'fake it till make it' approach?
Article is definitely mostly marketing, but if they managed to implement it (without using real people that doing dirty job filtering/parsing conversation) then this company is miles ahead of all other AI oriented companies including Google, Microsoft, OpenAI etc.
Honestly I'm quite sceptical of these claims, but I joined waiting list so.. lets see.
I wish people would use "confabulation" over "hallucination". Its just a better term for it
Heh, TIL.
Even StableDiffusion I wanna characterize more as "visual confabulation"
Edit: time will dictate the extent to which I support your view being clarified and discussed predominantly
I assume if you’re privacy conscious you should stay well away?