Possibly. You could make a notebook for every creator and try to distill their “essence” into a profile, then put all the profiles together. NotebookLM is great because it lets you control the sources, but it only lets you put 50 sources per notebook in the free version, and 300 in the premium version. So making a PVK bot would require a lot more work. Other AI’s can grab stuff from the open internet, but what they gather is always a crapshoot based on your prompting.
That is a detailed and thoughtful proposition. My initial reaction was that AI/LLM's aren't really thinking machines. So they can't do what you propose.
But if you're thinking that as AI technology advances that it could be programmed and trained to do what you suggested. I see the possibilities could be beneficial.
There is a ton of source material that I'm ignorant of. So I am probably missing a lot of things that back up and support your proposal.
It was interesting to read and ponder upon. Thank you for writing this.
What Elan Barenholtz work studying LLM's revealed is that the way we use language is basically identical. It doesn't mean that an LLM is really like a human, because we have embodied senses, but it does mean that the way we use language reveals a kind of "map" of our worldview. You can tell the difference when you talk to someone who just says the same things over and over again, versus people who really explore new ideas and like to expand their vocabularies, and that reveals something more useful than something like credentials.
Two people could both have PhD's, but one of them is very open, really understands the concepts behind their study, and can probably make analogies to other fields of study, while the other person just regurgitates the things they have memorized. If you study their use of language enough, you can tell the difference, but it takes a long time through conversation. AI can discover the difference much quicker. So the idea would be to kind of gauge where people are at, maybe push them to grow if possible, but also match people with others who share a good percentage of similarity in their thinking.
Not a hundred percent, but if two people are maybe 70% in alignment, there is a good chance that they can both learn to understand the other person's views where they don't line up. Then they both grow and become more capable of having better conversations with new people in the future. As far as I can tell, it's pretty feasible.
I wonder if that "use language" identically. I have seen and the agent admits they struggle to live in two languages at once. I also include the idea of language with multiple simultaneous meanings. Best AI can do so far is add multilingual Input without depth. More performative, when the human uses language transformationaly?
A other string for you is oxytocin aa an anti tribal hormone. By using this to show what we share, it activates this pro social perspective of the other?
This paragraph stood out to me: 'Through the AI intermediary, the user was able to look past her abrasive “indestructible map” to understand the sincere, vulnerable motivations driving her worldview: a profound hatred of sin, a rejection of cheap theological loopholes, and a desire for an “erotic ordeal of truth”. By privately translating her epistemological framework, the AI allowed the user to find unexpected cognitive overlap and helped draft a charitable, non-dismissive strategy for potential engagement.' This is a brilliant example of how AI can be used to aid in social coordination, increasing the probability that necessarily "surprising encounter" (which helps us avoid Deleuzian "capture") is far more likely to prove enriching and "convivial" versus overwhelming in a way that leads to exit and withdraw. Work and thinking like this is, for me, of the greatest importance in our current moment. Well done!
Fantastic, absolutely fantastic. The SCM is so important to me, and this idea of using AI gives me a lot of hope for AI to prove convivial as Illich stressed. Beautifully done!
I wonder if we could get this built into thislittlecorner.net as a feature.
Possibly. You could make a notebook for every creator and try to distill their “essence” into a profile, then put all the profiles together. NotebookLM is great because it lets you control the sources, but it only lets you put 50 sources per notebook in the free version, and 300 in the premium version. So making a PVK bot would require a lot more work. Other AI’s can grab stuff from the open internet, but what they gather is always a crapshoot based on your prompting.
That is a detailed and thoughtful proposition. My initial reaction was that AI/LLM's aren't really thinking machines. So they can't do what you propose.
But if you're thinking that as AI technology advances that it could be programmed and trained to do what you suggested. I see the possibilities could be beneficial.
There is a ton of source material that I'm ignorant of. So I am probably missing a lot of things that back up and support your proposal.
It was interesting to read and ponder upon. Thank you for writing this.
What Elan Barenholtz work studying LLM's revealed is that the way we use language is basically identical. It doesn't mean that an LLM is really like a human, because we have embodied senses, but it does mean that the way we use language reveals a kind of "map" of our worldview. You can tell the difference when you talk to someone who just says the same things over and over again, versus people who really explore new ideas and like to expand their vocabularies, and that reveals something more useful than something like credentials.
Two people could both have PhD's, but one of them is very open, really understands the concepts behind their study, and can probably make analogies to other fields of study, while the other person just regurgitates the things they have memorized. If you study their use of language enough, you can tell the difference, but it takes a long time through conversation. AI can discover the difference much quicker. So the idea would be to kind of gauge where people are at, maybe push them to grow if possible, but also match people with others who share a good percentage of similarity in their thinking.
Not a hundred percent, but if two people are maybe 70% in alignment, there is a good chance that they can both learn to understand the other person's views where they don't line up. Then they both grow and become more capable of having better conversations with new people in the future. As far as I can tell, it's pretty feasible.
I wonder if that "use language" identically. I have seen and the agent admits they struggle to live in two languages at once. I also include the idea of language with multiple simultaneous meanings. Best AI can do so far is add multilingual Input without depth. More performative, when the human uses language transformationaly?
A other string for you is oxytocin aa an anti tribal hormone. By using this to show what we share, it activates this pro social perspective of the other?
Excellent. Technology often feels like a threat until we find a place to use a tool to augment ourselves and see the potential?
This paragraph stood out to me: 'Through the AI intermediary, the user was able to look past her abrasive “indestructible map” to understand the sincere, vulnerable motivations driving her worldview: a profound hatred of sin, a rejection of cheap theological loopholes, and a desire for an “erotic ordeal of truth”. By privately translating her epistemological framework, the AI allowed the user to find unexpected cognitive overlap and helped draft a charitable, non-dismissive strategy for potential engagement.' This is a brilliant example of how AI can be used to aid in social coordination, increasing the probability that necessarily "surprising encounter" (which helps us avoid Deleuzian "capture") is far more likely to prove enriching and "convivial" versus overwhelming in a way that leads to exit and withdraw. Work and thinking like this is, for me, of the greatest importance in our current moment. Well done!
Yes, now we just need the AI to become the friend willing to pass notes to person we are too afraid to approach directly.
Fantastic, absolutely fantastic. The SCM is so important to me, and this idea of using AI gives me a lot of hope for AI to prove convivial as Illich stressed. Beautifully done!