I think LLMs don't yet have a good enough model of human behaviour to suggest potential mates no matter how many 10,000s of tokens are provided as a prompt. Maybe GPT-6 or 7 might?
Personally I think people are not so variable that there is anything to be gained from a pool of more than a few thousand nearby young adults. Call me old fashioned. I'd like to see a return to social norms favouring in-person interactions in social/project focused group activities. Sure there will be people with extreme aspects of their personality that can't be matched, though by definition they will be extremely rare and likely could be provisionally serviced by 'exclusive' non-location based (digital) matching. Probably applies to casual sex too.
It is nice to see that someone has even lower expectations of AI than I do.
However, I have a hunch that this is the sort of probabilistic statistics that AI is very good at.
Of course, I am sort of biased, having found what feels like a one in a million match. And this match would have been difficult to find on a traditional dating app since our interests are not very similar but rather complimentary. Such things are not that easy to put down on a form but an AI with plenty of data on what has worked before should probably be able to pull it off.
I think you have a good point. For most people, just going back a few decades would probably improve things greatly. But I think that ironically, an AI could be the easiest way to achieve that. It could narrow down the pool of candidates to the local and similar-minded, like things naturally were in the pre-internet times. Somehow blocking out those millions of candidates is one of the most important tasks.
I'm feeling a bit unfairly depicted here. I didn't think the idea to allow people to pay for casual sex was that bad. I don't think it would solve the matching problem between males and females, because I don't think most people would like to use such options. Still, I find the idea somewhat sympathetic, because it allows people to be honest. A man thinking "I would really like to know how having sex with her would be like" can be completely honest with that and send an offer (if the potential recipient accepts payments). That is more sympathetic than the alternative: Feign interest in her personality better than all his competitors.
In NZ prostitution is legal. I know of the pretty young adult daughter of an acquaintance of an acquaintance who has chosen to work as an prostitute. She's a matter-of fact person and seems fairly public about it. This is in an upper middle class suburb rather than the central city where I understand most prostitutes are from out of town (or further afield).
For someone with the right kind of personality it can probably be a clever choice. But I guess it is important to have a good pension plan. Women in that business decline as fast as professional footballers.
She might know something that they don't know. For example, she might be highly sexual. I have studied the phenomenon of voluntary middle-class sex workers as much as possible and in the few cases I could find that was an important factor.
Highly sexual women is a very understudied (and taboo) topic. I'd love to hear what you concluded from your study.
There was one Danish book that interviewed women who frequented sex clubs ("Blot til lyst" by Carsten Graff), which gave an interesting sociological perspective on highly sexual women that you don't otherwise hear much about.
It was interesting to read how a few of them at one point grouped together and started a prostitution service. Not because they needed the money (they all had well-paying jobs) nor needing sexual partners (which they had plenty of at the clubs), but simply because they wanted to have fun and found it a turn-on to play prostitutes.
Thank you for the book tip, I will try to get hold of a copy. I had never heard of Carsten Graff, although he seems pretty well-known in Denmark. Denmark and Sweden really are different duck ponds.
Yes, highly sexual women is an understudied topic. The only related topic that is even more understudied is averagely sexual women: There are many more biography-style books about highly sexual women than about normally sexual women. Maybe highly sexual and less sexual men is an understudied topic too, because I find even fewer books about them.
If some women actually enjoy selling sex, it is not the strangest sexual kink someone in this world has. I mean, people are turned on by such a variety of things, some of them obviously self-destructive. In that light it would be strange if absolutely no one enjoyed sex work.
The AI could have a conversation with you. Something like Pi (https://inflection.ai/) could start by asking you to describe your friends, and then get into a longer conversation that would deepen those descriptions in order to elaborate on your interests and values. Then it could look at some past data consisting of conversations with happy couples to decide who might be a good match. Unlike other matchmaking apps, this would not use self-reporting and quizzes. Basing its decisions on conversations, it would be more like a friend who knows two people and says, "Hey, I should introduce them."
Sorry but I can't go with your thinking here. For the great majority of men on a dating app - and irrespective of whether they are looking for casual sex or a long-term partner - the "good match" is the prettiest woman they can manage to get a response from. There can be few things more deadening in the world of dating than the person who shares your "interests and values" but you do not fancy.
This is what I am imagining. Dating apps are limited by the number of individuals you are able to strike up conversations with. You are supposed to have fairly similar conversations with a lot of people, just to be able to make an assessment of them. By talking to an AI instead you can concentrate your resources on a single conversation and let the AI do the assessing for you. The practical effect should be that you can assess tens of thousands of prospective partners rather than the mere tens you would have been able to have a conversation with yourself.
By the way, the description of the AI as a mutual friend who introduces people to each other is just great. Had I thought of it myself I would have put it in the article.
To go one step further, instead of having an active conversation with the AI, give it access to your browsing data, phone conversations, etc, so that it can make matches based on who you are vs how you portray yourself.
I agree that self-portrayal is a problem, but I prefer a conversation as the solution. The conversation has to be more nuanced than "describe yourself." I like the idea of starting with having someone describe their friends
One thing I wonder is: What are all psychologists doing? There should be psychological research that says what information gives an honest description of people, what similarities and differences tend to create stable and happy families etc. Finding out what makes people like each other could be a very important job for psychology, but I can see little such research.
There should even be research trying to answer the question: Is honesty even needed, or can people be rather successfully matched based on their white lies?
Maybe underutilization of western brains in romance matters/mate selection has been selected for.
I think LLMs don't yet have a good enough model of human behaviour to suggest potential mates no matter how many 10,000s of tokens are provided as a prompt. Maybe GPT-6 or 7 might?
Personally I think people are not so variable that there is anything to be gained from a pool of more than a few thousand nearby young adults. Call me old fashioned. I'd like to see a return to social norms favouring in-person interactions in social/project focused group activities. Sure there will be people with extreme aspects of their personality that can't be matched, though by definition they will be extremely rare and likely could be provisionally serviced by 'exclusive' non-location based (digital) matching. Probably applies to casual sex too.
It is nice to see that someone has even lower expectations of AI than I do.
However, I have a hunch that this is the sort of probabilistic statistics that AI is very good at.
Of course, I am sort of biased, having found what feels like a one in a million match. And this match would have been difficult to find on a traditional dating app since our interests are not very similar but rather complimentary. Such things are not that easy to put down on a form but an AI with plenty of data on what has worked before should probably be able to pull it off.
I think you have a good point. For most people, just going back a few decades would probably improve things greatly. But I think that ironically, an AI could be the easiest way to achieve that. It could narrow down the pool of candidates to the local and similar-minded, like things naturally were in the pre-internet times. Somehow blocking out those millions of candidates is one of the most important tasks.
I'm feeling a bit unfairly depicted here. I didn't think the idea to allow people to pay for casual sex was that bad. I don't think it would solve the matching problem between males and females, because I don't think most people would like to use such options. Still, I find the idea somewhat sympathetic, because it allows people to be honest. A man thinking "I would really like to know how having sex with her would be like" can be completely honest with that and send an offer (if the potential recipient accepts payments). That is more sympathetic than the alternative: Feign interest in her personality better than all his competitors.
In NZ prostitution is legal. I know of the pretty young adult daughter of an acquaintance of an acquaintance who has chosen to work as an prostitute. She's a matter-of fact person and seems fairly public about it. This is in an upper middle class suburb rather than the central city where I understand most prostitutes are from out of town (or further afield).
For someone with the right kind of personality it can probably be a clever choice. But I guess it is important to have a good pension plan. Women in that business decline as fast as professional footballers.
Yes, though I gathered her parents were somewhat nonplussed. Its not that she is highly social.
She might know something that they don't know. For example, she might be highly sexual. I have studied the phenomenon of voluntary middle-class sex workers as much as possible and in the few cases I could find that was an important factor.
Highly sexual women is a very understudied (and taboo) topic. I'd love to hear what you concluded from your study.
There was one Danish book that interviewed women who frequented sex clubs ("Blot til lyst" by Carsten Graff), which gave an interesting sociological perspective on highly sexual women that you don't otherwise hear much about.
It was interesting to read how a few of them at one point grouped together and started a prostitution service. Not because they needed the money (they all had well-paying jobs) nor needing sexual partners (which they had plenty of at the clubs), but simply because they wanted to have fun and found it a turn-on to play prostitutes.
Thank you for the book tip, I will try to get hold of a copy. I had never heard of Carsten Graff, although he seems pretty well-known in Denmark. Denmark and Sweden really are different duck ponds.
Yes, highly sexual women is an understudied topic. The only related topic that is even more understudied is averagely sexual women: There are many more biography-style books about highly sexual women than about normally sexual women. Maybe highly sexual and less sexual men is an understudied topic too, because I find even fewer books about them.
If some women actually enjoy selling sex, it is not the strangest sexual kink someone in this world has. I mean, people are turned on by such a variety of things, some of them obviously self-destructive. In that light it would be strange if absolutely no one enjoyed sex work.
The AI could have a conversation with you. Something like Pi (https://inflection.ai/) could start by asking you to describe your friends, and then get into a longer conversation that would deepen those descriptions in order to elaborate on your interests and values. Then it could look at some past data consisting of conversations with happy couples to decide who might be a good match. Unlike other matchmaking apps, this would not use self-reporting and quizzes. Basing its decisions on conversations, it would be more like a friend who knows two people and says, "Hey, I should introduce them."
Sorry but I can't go with your thinking here. For the great majority of men on a dating app - and irrespective of whether they are looking for casual sex or a long-term partner - the "good match" is the prettiest woman they can manage to get a response from. There can be few things more deadening in the world of dating than the person who shares your "interests and values" but you do not fancy.
This is what I am imagining. Dating apps are limited by the number of individuals you are able to strike up conversations with. You are supposed to have fairly similar conversations with a lot of people, just to be able to make an assessment of them. By talking to an AI instead you can concentrate your resources on a single conversation and let the AI do the assessing for you. The practical effect should be that you can assess tens of thousands of prospective partners rather than the mere tens you would have been able to have a conversation with yourself.
By the way, the description of the AI as a mutual friend who introduces people to each other is just great. Had I thought of it myself I would have put it in the article.
The Mutual Friend sounds like a good name for our app ;)
To go one step further, instead of having an active conversation with the AI, give it access to your browsing data, phone conversations, etc, so that it can make matches based on who you are vs how you portray yourself.
Are you serious?
I agree that self-portrayal is a problem, but I prefer a conversation as the solution. The conversation has to be more nuanced than "describe yourself." I like the idea of starting with having someone describe their friends
One thing I wonder is: What are all psychologists doing? There should be psychological research that says what information gives an honest description of people, what similarities and differences tend to create stable and happy families etc. Finding out what makes people like each other could be a very important job for psychology, but I can see little such research.
There should even be research trying to answer the question: Is honesty even needed, or can people be rather successfully matched based on their white lies?