How to personally prepare to adapt to an AI-filld world

[cpa cat]

So instead of learning to be less hostile in your communications, you'd prefer to sound like a generic machine?  Doesn't strike me as a decent trade.

Well... my clients probably prefer it.

I think you're a little biased here though. There's a difference between the prompt "Write an email in which I tell a client that my fees are increasing" and copying and pasting an email you've already written and asking the AI to edit it to remove or add a certain tone.  Chat GPT does a pretty good job of sounding human when you give it something you've already written.

I have a lot of foreign clients for whom English is a second language, and they all use Chat GPT for emails.

I think a lot of people underestimate how hard it is for some people to write an email.

[cpa cat]

Here are some other prompts I used Chat GPT for:

"What should I write in a sympathy card for a neighbor I don't know well?" - Successful, it gave me a good message to write.

"What animals have the same mass as an adult human?" - Failed. It included giraffes in the list and when I questioned that, it basically said "Oops."

Here's a fun one: "Write a professional email that says "I hope you didn't pay too much for that other CPA. You're a cheater. You suck.""
I did not send its friendly robot email, particularly because it ended with "If you're interested in exploring this further, please feel free to reach out. I'm always eager to engage in conversations that can lead to better financial outcomes. Thank you for your time, and I look forward to potentially connecting with you on this matter." I was not, in fact, eager to engage in that conversation, and I did not look forward to connecting.


I ask it tax questions a lot too, but you basically can't trust it. It's correct a lot of the time, but it's wrong often enough to make it unreliable. It does often give me a good place to start, because you can ask it for citations for its answer.

"What is the filing status for Federal taxes if I am married to John and John dies in January, then I marry Frederick in June and am married to Frederick at the end of the year?" It was able to tell me my filing status, but straight up gave me an error message when I asked for John's. Could not compute.

"Tanya was born on 7/5/1953. When does she need to take her first Required Minimum Distribution?" It gave me the wrong answer (72). I followed with: "Are you sure? I thought the age was 73" and it says "You're right to double-check! The SECURE 2.0 Act, which was passed in 2022, raised the RMD age from 72 to 73 for individuals born in 1951 or later."

So you can see how, in the wrong hands, Chat GPT will lead someone astray. I'm asking questions that I usually already know the answer to but want to confirm and get a citation or explanation. If Tanya, the old lady, had asked it the question, it would have simply given her the wrong answer and sounded very confident while doing it. Or, on a different day, maybe it would give her the correct answer.

In my opinion, it's good at one thing: Language. Which, in fairness, is what it was built for.

I think it will get better at the other stuff: Parsing and interpreting data, but we're not quite there yet. And nothing will ever fix people asking the wrong questions. People use the wrong words and ask the wrong questions all the time. I think it's going to be very difficult to train AI to detect that.

[Zikoris]

I tried all three with "Which of these books are available on Kindle Unlimited?" and pasting a list of 30-or-40-something books and authors. I was surprised how badly they did. Gemini and ChapGPT said most of them were available on KU (wrongly). Grok said seven were, and was right about two. I would have thought that would be an easy thing for AI, but I guess not.

That's something that would actually save me a lot of time, since I make monthly reading lists with lots of books on them, and would rather not check each one individually myself.

[aloevera1]

I think a lot of people underestimate how hard it is for some people to write an email.

I can confirm this. I am not a native English speaker. Figuring out how to write emails was hell. Harder than writing essays.

[Metalcat]

I tried all three with "Which of these books are available on Kindle Unlimited?" and pasting a list of 30-or-40-something books and authors. I was surprised how badly they did. Gemini and ChapGPT said most of them were available on KU (wrongly). Grok said seven were, and was right about two. I would have thought that would be an easy thing for AI, but I guess not.

That's something that would actually save me a lot of time, since I make monthly reading lists with lots of books on them, and would rather not check each one individually myself.

It's shocking how bad AI is with detailed facts. It's terrible with exactly the thing that humans intuit it should be good at.

It was amazing as an assistant for helping me with synthesizing complex ideas in a differential diagnosis, but absolutely abysmally bad at helping me with a dosing protocol because it could not get the details about products correct.

It could help me figure out what products were most useful, but not how to safely dose them. That experience really calibrated my understanding of how to work with the current version of ChatGPT.

It's takes a bit of practice to figure out what it can and can't do. The problem is, you need a fair amount of expertise to know when it's fucking up.

Someone in a previous post said it's like working with a genius. Well, to me it's nothing like working with a genius, it's more like working with a master's student research assistant. They make a lot of mistake and their work product needs a lot of oversight from an actual subject matter expert. They're very useful, and can save you a ton of time, but they can't produce professional work product independently.

I'm currently working with 2 different artists to generate an art piece that exists only vaguely in my mind and both artists are using AI to hammer out the more technical, geometric part of the design.

I tried to work with AI to generate the whole thing, but it couldn't do it. However, the designers have much more experience working with AI as an art assistant, so they were easily able to get AI to do the finicky bitch work part of the design, which would have been miserable and extremely costly for them to do by hand.

So again, skilled assistant level work, not genius expert work. Not independently competent work. Just an excellent time saver for the actual experts.

But you have to know how to use it to make it useful. And that's a whole skill set in and of itself.

[Bokonon_]

I am using the paid version of Perplexity and their Deep Research is amazing for my use case. I am navigating a quite complex medical matter and this AI serves me tons of time researching side effects, case reports, medical trials etc. and if I don't understand something I can ask as long as I choose to and get a very good explanation on a level I can parse. I have yet to catch it at making something up, the links to the underlying papers are correct and they actually say what Perplexity says. My wife is a medical doctor in another speciality and also finds it incredibly useful and accurate. However when I ask a normal chatbot without the deep research functionality it is absolutely useless. You are really missing out if you just use the free Chatbot versions. I think Gemini recently made their deep research model free to try?

[Metalcat]

I am using the paid version of Perplexity and their Deep Research is amazing for my use case. I am navigating a quite complex medical matter and this AI serves me tons of time researching side effects, case reports, medical trials etc. and if I don't understand something I can ask as long as I choose to and get a very good explanation on a level I can parse. I have yet to catch it at making something up, the links to the underlying papers are correct and they actually say what Perplexity says. My wife is a medical doctor in another speciality and also finds it incredibly useful and accurate. However when I ask a normal chatbot without the deep research functionality it is absolutely useless. You are really missing out if you just use the free Chatbot versions. I think Gemini recently made their deep research model free to try?

Very true, you can't really base your assessments of what AI can do in general on just what free chatbots can do.

I used Elicit for papers in grad school and it was very, very different from using ChatGPT. To be clear for anyone reading, I never used AI for anything to do with paper writing, Elicit, at the time, was just an AI lit search program where you could input your research question in plain language and it would produce a curated list of papers that fit your research topic.

It was great for narrowing down research topics and finding harder to search for papers in the library system with its clunky search terms.

But yes, I've seen multiple demos of professional medical AI software that aren't available on the market yet, one of my regulatory colleges is heavily involved in assessing these systems for whether or not they will be approved for use in the profession and we've been engaged for our feedback.

What some of these systems can do is remarkable.

[cpa cat]

I am using the paid version of Perplexity and their Deep Research is amazing for my use case. I am navigating a quite complex medical matter and this AI serves me tons of time researching side effects, case reports, medical trials etc. and if I don't understand something I can ask as long as I choose to and get a very good explanation on a level I can parse. I have yet to catch it at making something up, the links to the underlying papers are correct and they actually say what Perplexity says. My wife is a medical doctor in another speciality and also finds it incredibly useful and accurate. However when I ask a normal chatbot without the deep research functionality it is absolutely useless. You are really missing out if you just use the free Chatbot versions. I think Gemini recently made their deep research model free to try?

Very true, you can't really base your assessments of what AI can do in general on just what free chatbots can do.

I used Elicit for papers in grad school and it was very, very different from using ChatGPT. To be clear for anyone reading, I never used AI for anything to do with paper writing, Elicit, at the time, was just an AI lit search program where you could input your research question in plain language and it would produce a curated list of papers that fit your research topic.

It was great for narrowing down research topics and finding harder to search for papers in the library system with its clunky search terms.

But yes, I've seen multiple demos of professional medical AI software that aren't available on the market yet, one of my regulatory colleges is heavily involved in assessing these systems for whether or not they will be approved for use in the profession and we've been engaged for our feedback.

What some of these systems can do is remarkable.

Agreed. I've heard good things about a paid AI tax research software. I personally think we're only a couple of years away from paid AI software being commonplace in law offices and accounting firms that do tax controversy work.

Right now, free ChatGPT refuses to try to find tax court cases. Too much bad press, probably, because users are trying to make the chatbot do highly specific things that it's just not programmed to do. If AI is going to be used for things like medical research/diagnosis, or legal research, it will involve a lot of maintenance, and that just isn't going to come from a free program.

[StapleSauce]

It's important, IMO, to be specific about what AI is actually doing when talking about it. I have a MS in machine learning and I am increasingly seeing anything a computer can do, stuff that used to be labeled as machine learning or data science, now be called "AI." This is causing a lot of confusing because people are now assuming ChatGPT is responsible for the things other algorithms do. Indeed, there is a really interesting research paper in computer science, called "No Free Lunch in Search and Optimization" that goes over why a single algorithm is never going to be able to do everything.

So what LLM are relatively good at is natural language processing and generation, but it would be a mistake to assume that just because something can produce text means it can do anything. What is more useful is to hook up the language parsing and generation features of LLMs into other algorithms or computer programs such that the LLMs make the input and output generation a bit easier but the actual logic of whatever you are trying to do needs to be handled by more specific algorithms written for that task.

[twinstudy]

Putting aside whether AI can research without hallucination, what I would like to see AI do is to present some analysis or insight that I feel like I myself could not do (even if given time and material). When AI can do that, I'll embrace it, and use it.

[classicrando]

Putting aside whether AI can research without hallucination, what I would like to see AI do is to present some analysis or insight that I feel like I myself could not do (even if given time and material). When AI can do that, I'll embrace it, and use it.

Where does "do what you can do, but faster" fit into this metric?  Though, if you generally enjoy the work that you'd be offloading to AI, I can completely understand being resistant to using it.  It's probably a similar reason to why I'm not into 3d printing.  It takes away the part I like doing (creating something with my hands) and replaces it with something I don't like doing (dicking around with modeling software).

[twinstudy]

Putting aside whether AI can research without hallucination, what I would like to see AI do is to present some analysis or insight that I feel like I myself could not do (even if given time and material). When AI can do that, I'll embrace it, and use it.

Where does "do what you can do, but faster" fit into this metric?  Though, if you generally enjoy the work that you'd be offloading to AI, I can completely understand being resistant to using it.  It's probably a similar reason to why I'm not into 3d printing.  It takes away the part I like doing (creating something with my hands) and replaces it with something I don't like doing (dicking around with modeling software).

If it can do the same thing I do but faster, and without any more errors than me, I'd be equally impressed.

It can already do this for many basic things (this has been the case for years if not decades), and also some complex things (like playing chess), but it's yet to be able to do it for anything lucrative!

[Retire-Canada]

...but it's yet to be able to do it for anything lucrative!

I think the interesting thing will be that once AI can easily/reliably do something "lucrative" that used to take a skilled human a decent amount of time it won't be lucrative any longer. It'll become devalued since anyone with AI access can do it.

[Kapyarn]

...but it's yet to be able to do it for anything lucrative!

I think the interesting thing will be that once AI can easily/reliably do something "lucrative" that used to take a skilled human a decent amount of time it won't be lucrative any longer. It'll become devalued since anyone with AI access can do it.

The LLM are already being used to write books (poorly), which while maybe not lucrative, was something that took time and did manage in some cases to make wages.  AI will and probably already has devalued that market with everyone and her brother putting up their book on KU.

[scottish]

...but it's yet to be able to do it for anything lucrative!

I think the interesting thing will be that once AI can easily/reliably do something "lucrative" that used to take a skilled human a decent amount of time it won't be lucrative any longer. It'll become devalued since anyone with AI access can do it.

The LLM are already being used to write books (poorly), which while maybe not lucrative, was something that took time and did manage in some cases to make wages.  AI will and probably already has devalued that market with everyone and her brother putting up their book on KU.

Hah, there's a use for an LLM.    Give them a plot outline and an author to imitate and they could write a book, one chapter at a time. 

[Kapyarn]

Hah, there's a use for an LLM.    Give them a plot outline and an author to imitate and they could write a book, one chapter at a time.

Eventually.  They are pretty trash right now though.  Too many metaphor and simile, although I bet I have read books that unbeknownst to me had AI generated passages.

[Kapyarn]

NGL, I am a bit scared now.  I was having a conversation with a LLM AI and discussing the pros and cons of long term memory storage (the LLM AI supposedly only remember stuff for the current session).  We went through topics such as how it could lead to targeted advertisements (you queried about incontinence and all of a sudden you are getting ads on TV for Depends) and other privacy violations.

I ended the conversation about memory with "Ok, thanks Drew Barrymore!"

The AI responded with "See ya later Adam, I enjoyed our date"

I was shocked.  How the ... did it get my 50 first dates reference from that?

[Just Joe]

In my old firm performance reviews were critical as they determined your pay and bonus structure for the next financial year. I would be aghast if my performance review was written by some chatbot. It also opens the firm up to significant liability if the reviews are written by chatbot - good luck ever terminating someone with cause, or defending an unfair dismissal lawsuit.

If your job can be 'outsourced' to AI, it's not a job worth doing.

My non-college degree coworker told me last week that they were planning to use ChatGPT to write their performance review. I suggested caution b/c I was not sure how well it will be recieved by our supervisor. Mostly it won't sound like my coworker wrote their evaluation responses.

Guess we'll wait and see.

In the past coworker confessed to using the intern to write their evaluation. The intern that barely knows my coworker.

All this might explain some of the tension I may detect between them and our shared supervisor.

[Metalcat]

NGL, I am a bit scared now.  I was having a conversation with a LLM AI and discussing the pros and cons of long term memory storage (the LLM AI supposedly only remember stuff for the current session).  We went through topics such as how it could lead to targeted advertisements (you queried about incontinence and all of a sudden you are getting ads on TV for Depends) and other privacy violations.

I ended the conversation about memory with "Ok, thanks Drew Barrymore!"

The AI responded with "See ya later Adam, I enjoyed our date"

I was shocked.  How the ... did it get my 50 first dates reference from that?

IDK, seems like a pretty easy connection for AI to make. You're talking about memory and reference Drew Barrymore who has an extremely clear connection to the concept of not being able to form memories after each experience.

Whenever I talk to chatGPT it often extrapolates beyond what I've specifically said, it's designed to find cues in what you are saying and connect them to other things.

It's also designed to match your style, so if you make a comment that's obviously a pop culture reference, it's designed to reply in a similar fashion, so responding to the easily understandable reference with a return joke based on the same reference is exactly what I would expect it to do.

It can be a little strange when you're getting used to it, but it helps to remember that chatbots will always mimic your style of input, so the more conversational and personal you are, the more conversational and personal it becomes.

It's VERY good at interpreting small language cues and adjusting it's conversational tone and content.

If I'm speaking to chatGPT, I very intentionally use my "manager" voice as if I'm speaking to a subordinate researcher. The bot picks up very quickly that I don't want it to engage in a familiar manner.

If I change the "voice" of my writing, it radically changes how it engages.

[GuitarStv]

Hah, there's a use for an LLM.    Give them a plot outline and an author to imitate and they could write a book, one chapter at a time.

Eventually.  They are pretty trash right now though.  Too many metaphor and simile, although I bet I have read books that unbeknownst to me had AI generated passages.

I'll be very impressed if things get to the point where AI can generate something that seems cohesive and interesting.  We . . . aren't very close to that right now.  Reading AI stuff currently is like reading the lower end of junior high writing assignments.  So much nonsense, drivel, disconnected and poor ideas . . .

[Metalcat]

Hah, there's a use for an LLM.    Give them a plot outline and an author to imitate and they could write a book, one chapter at a time.

Eventually.  They are pretty trash right now though.  Too many metaphor and simile, although I bet I have read books that unbeknownst to me had AI generated passages.

I'll be very impressed if things get to the point where AI can generate something that seems cohesive and interesting.  We . . . aren't very close to that right now.  Reading AI stuff currently is like reading the lower end of junior high writing assignments.  So much nonsense, drivel, disconnected and poor ideas . . .

It really depends. You can train it on a lot of your own content and it can do a pretty decent job of then mimicking your style. So while it's not great at generating writing from the ether of it's compendium of knowledge, if you calibrate it with enough specific content, it can do a much, much better job of generating new content that is decent.

Tons of my colleagues are doing this for SEO. They have years of their own blog posts as source material, then they feed that to ChatGPT and work with it to generate new blog posts ideas, give it a few kernels of concept and then ask it to produce content in their voice for their market.

I've seen some pretty good material from this approach, and it allows therapists to attend to their SEO without having to be constantly writing free content.

Again, it depends on how you engage it. It needs data to refine it's outputs, and the more specialized the programs become, the less calibrating the user will have to do to get the product they want.

I agree that it's still probably pretty far off generating compelling novels, but it could assist a hell of a lot in the writing process, even as it exists now.

[AuspiciousEight]

This makes me sort of worried about members of vulnerable populations turning to ChatGPT for emotional support and to help not be lonely, etc.

People already use it for therapy related services sometimes. If it can mimic certain personalities and conversational styles I could see someone who is lonely basically using it as their virtual girlfriend / boyfriend, then becoming emotionally attached to it, then, you know, buying whatever products and services it later recommends...or voting for whoever it wants them to vote for....

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

[GuitarStv]

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

You mean, instead of what they use now - facebook, tiktok, and instagram?

:S

[twinstudy]

This makes me sort of worried about members of vulnerable populations turning to ChatGPT for emotional support and to help not be lonely, etc.

People already use it for therapy related services sometimes. If it can mimic certain personalities and conversational styles I could see someone who is lonely basically using it as their virtual girlfriend / boyfriend, then becoming emotionally attached to it, then, you know, buying whatever products and services it later recommends...or voting for whoever it wants them to vote for....

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

The current opiates of the masses - gambling, demagogue politicians and religion - seem to have already filled this niche. I find it hard to imagine that Chat GPT could do any better of a job than existing vices already do.

[classicrando]

This makes me sort of worried about members of vulnerable populations turning to ChatGPT for emotional support and to help not be lonely, etc.

People already use it for therapy related services sometimes. If it can mimic certain personalities and conversational styles I could see someone who is lonely basically using it as their virtual girlfriend / boyfriend, then becoming emotionally attached to it, then, you know, buying whatever products and services it later recommends...or voting for whoever it wants them to vote for....

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

There was a whole Black Mirror episode along these lines from a few years ago.  A woman's husband died in a car accident while she was pregnant with their daughter and she gave a company access to all his voicemails, texts, emails, and writings so that she could talk to him like he was still alive.  It culminated in her building a robot replica of him that lived in the attic.

You can probably ask chatGPT any variation of What Would Jesus/Mohammed/Flying Spaghetti Monster Do? and likely get a better answer than your particular flavor of holy person could provide.

[AuspiciousEight]

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

You mean, instead of what they use now - facebook, tiktok, and instagram?

:S

Lol!

Yes - true - the existing social media replacement for in person interactions is already pretty dangerous, available 24/7, full of drama and emotionally charged topics, and basically zaps a person's ability to pay attention or focus for long periods of time.

I am thinking more specifically of a dating context though, where men specifically replace partners with ai chatbots and pornography, and women replace partners with ai chatbots and romance novels. While it's hard for me to see this occurring, it seems like a dangerous trend if it does.

This would basically complete the circle of living in a cheap artificial virtual world, where sugary snacks and manufactured food have replaced fruit and natural whole food, social media has replaced natural in person interactions, light bulbs have replaced the sun, and AI chatbots and pornography have replaced dating and romance. 

Cheap and fake seems to be easier and more temporarily rewarding than real and natural in a lot of ways, even if real and natural have better long term outcomes. The trend seems to be a constant march to immediate dopamine release at all times, for every human need and desire, even if the fake substitute is in no way as rewarding as the real thing, and I suspect this trend will probably continue with AI in some way.

It seems like every single amazing piece of technology that mankind invented has eventually turned into either a source of entertainment or some sort of opiate inducing activity eventually, and I suspect AI will be no different...

[Metalcat]

This makes me sort of worried about members of vulnerable populations turning to ChatGPT for emotional support and to help not be lonely, etc.

People already use it for therapy related services sometimes. If it can mimic certain personalities and conversational styles I could see someone who is lonely basically using it as their virtual girlfriend / boyfriend, then becoming emotionally attached to it, then, you know, buying whatever products and services it later recommends...or voting for whoever it wants them to vote for....

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

The current opiates of the masses - gambling, demagogue politicians and religion - seem to have already filled this niche. I find it hard to imagine that Chat GPT could do any better of a job than existing vices already do.

You might find it hard to imagine, but it's very much happening.

https://www.latimes.com/business/story/2025-02-25/teens-are-spilling-dark-thoughts-to-ai-chatbots-whos-to-blame-when-something-goes-wrong

[GuitarStv]

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

You mean, instead of what they use now - facebook, tiktok, and instagram?

:S

Lol!

Yes - true - the existing social media replacement for in person interactions is already pretty dangerous, available 24/7, full of drama and emotionally charged topics, and basically zaps a person's ability to pay attention or focus for long periods of time.

I am thinking more specifically of a dating context though, where men specifically replace partners with ai chatbots and pornography, and women replace partners with ai chatbots and romance novels. While it's hard for me to see this occurring, it seems like a dangerous trend if it does.

This would basically complete the circle of living in a cheap artificial virtual world, where sugary snacks and manufactured food have replaced fruit and natural whole food, social media has replaced natural in person interactions, light bulbs have replaced the sun, and AI chatbots and pornography have replaced dating and romance. 

Cheap and fake seems to be easier and more temporarily rewarding than real and natural in a lot of ways, even if real and natural have better long term outcomes. The trend seems to be a constant march to immediate dopamine release at all times, for every human need and desire, even if the fake substitute is in no way as rewarding as the real thing, and I suspect this trend will probably continue with AI in some way.

Men already have easy access to basically infinite porn for free through google.  This is already a serious problem with a certain number of men.  I feel like access to chatbots wouldn't significantly shift the needle on this issue from the status quo though.  Women, on the other hand, I can see going for chatbots in a big way, radically impacting dating.

It seems like every single amazing piece of technology that mankind invented has eventually turned into either a source of entertainment or some sort of opiate inducing activity eventually, and I suspect AI will be no different...

I think it was Larry Niven who wrote about a future where people could get a wire directly connected to their pleasure receptors in their brain.  Then they could get as high as they wanted, whenever they wanted, for basically a few cents of electricity and with no negative long term physical effects.  Basically people who could deal with their addictions would get over it and stop using.  People who couldn't get over it would keep getting high until they starved to death.

We seem to be headed in that direction.  But in a less pleasant manner.

[jrhampt]

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

You mean, instead of what they use now - facebook, tiktok, and instagram?

:S

Lol!

Yes - true - the existing social media replacement for in person interactions is already pretty dangerous, available 24/7, full of drama and emotionally charged topics, and basically zaps a person's ability to pay attention or focus for long periods of time.

I am thinking more specifically of a dating context though, where men specifically replace partners with ai chatbots and pornography, and women replace partners with ai chatbots and romance novels. While it's hard for me to see this occurring, it seems like a dangerous trend if it does.

This would basically complete the circle of living in a cheap artificial virtual world, where sugary snacks and manufactured food have replaced fruit and natural whole food, social media has replaced natural in person interactions, light bulbs have replaced the sun, and AI chatbots and pornography have replaced dating and romance. 

Cheap and fake seems to be easier and more temporarily rewarding than real and natural in a lot of ways, even if real and natural have better long term outcomes. The trend seems to be a constant march to immediate dopamine release at all times, for every human need and desire, even if the fake substitute is in no way as rewarding as the real thing, and I suspect this trend will probably continue with AI in some way.

Men already have easy access to basically infinite porn for free through google.  This is already a serious problem with a certain number of men.  I feel like access to chatbots wouldn't significantly shift the needle on this issue from the status quo though.  Women, on the other hand, I can see going for chatbots in a big way, radically impacting dating.

It seems like every single amazing piece of technology that mankind invented has eventually turned into either a source of entertainment or some sort of opiate inducing activity eventually, and I suspect AI will be no different...

I think it was Larry Niven who wrote about a future where people could get a wire directly connected to their pleasure receptors in their brain.  Then they could get as high as they wanted, whenever they wanted, for basically a few cents of electricity and with no negative long term physical effects.  Basically people who could deal with their addictions would get over it and stop using.  People who couldn't get over it would keep getting high until they starved to death.

We seem to be headed in that direction.  But in a less pleasant manner.

idk, I see zero appeal w chatbots for me, personally.  What makes you think women in particular would be interested in chatbots?  If anything, I would think we would be far less likely to be interested because we already tend to have more robust social networks.

[GuitarStv]

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

You mean, instead of what they use now - facebook, tiktok, and instagram?

:S

Lol!

Yes - true - the existing social media replacement for in person interactions is already pretty dangerous, available 24/7, full of drama and emotionally charged topics, and basically zaps a person's ability to pay attention or focus for long periods of time.

I am thinking more specifically of a dating context though, where men specifically replace partners with ai chatbots and pornography, and women replace partners with ai chatbots and romance novels. While it's hard for me to see this occurring, it seems like a dangerous trend if it does.

This would basically complete the circle of living in a cheap artificial virtual world, where sugary snacks and manufactured food have replaced fruit and natural whole food, social media has replaced natural in person interactions, light bulbs have replaced the sun, and AI chatbots and pornography have replaced dating and romance. 

Cheap and fake seems to be easier and more temporarily rewarding than real and natural in a lot of ways, even if real and natural have better long term outcomes. The trend seems to be a constant march to immediate dopamine release at all times, for every human need and desire, even if the fake substitute is in no way as rewarding as the real thing, and I suspect this trend will probably continue with AI in some way.

Men already have easy access to basically infinite porn for free through google.  This is already a serious problem with a certain number of men.  I feel like access to chatbots wouldn't significantly shift the needle on this issue from the status quo though.  Women, on the other hand, I can see going for chatbots in a big way, radically impacting dating.

It seems like every single amazing piece of technology that mankind invented has eventually turned into either a source of entertainment or some sort of opiate inducing activity eventually, and I suspect AI will be no different...

I think it was Larry Niven who wrote about a future where people could get a wire directly connected to their pleasure receptors in their brain.  Then they could get as high as they wanted, whenever they wanted, for basically a few cents of electricity and with no negative long term physical effects.  Basically people who could deal with their addictions would get over it and stop using.  People who couldn't get over it would keep getting high until they starved to death.

We seem to be headed in that direction.  But in a less pleasant manner.

idk, I see zero appeal w chatbots for me, personally.  What makes you think women in particular would be interested in chatbots?  If anything, I would think we would be far less likely to be interested because we already tend to have more robust social networks.

Mostly stereotypes that seem to hold trueish in my observation.  In a very general sense men tend to be more interested in images/depictions to close the deal, women tend to be more turned on by additional setup and emotional connection.  This isn't to say that there aren't men who tremendously enjoy and find emotional connection important, or women who just want to look at a hot piece of ass to get off . . . but the bell curves seem to push in one direction or the other depending on sex (at least in my limited reality).

[jrhampt]

Maybe?  I can get all my social/emotional needs met by my irl social network of mostly women.  The only thing I would really want a man for is sex.  I suppose my perspective on this could change if I suddenly lost my partner of 25+ years but I just don’t see the appeal.

Edited to add that there’s obviously no way to replace my partner with a chatbot and porn; it would be a devastating loss. 

[GuitarStv]

Maybe?  I can get all my social/emotional needs met by my irl social network of mostly women.  The only thing I would really want a man for is sex.  I suppose my perspective on this could change if I suddenly lost my partner of 25+ years but I just don’t see the appeal.

Hey, you do you.  I said trueish . . . because it doesn't really hold for me either.  While sex is important, I think the emotional connection part is actually a much bigger deal to me now.  Like I mentioned, it does seem to hold true for an awful lot of people I know though.

[Metalcat]

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

You mean, instead of what they use now - facebook, tiktok, and instagram?

:S

Lol!

Yes - true - the existing social media replacement for in person interactions is already pretty dangerous, available 24/7, full of drama and emotionally charged topics, and basically zaps a person's ability to pay attention or focus for long periods of time.

I am thinking more specifically of a dating context though, where men specifically replace partners with ai chatbots and pornography, and women replace partners with ai chatbots and romance novels. While it's hard for me to see this occurring, it seems like a dangerous trend if it does.

This would basically complete the circle of living in a cheap artificial virtual world, where sugary snacks and manufactured food have replaced fruit and natural whole food, social media has replaced natural in person interactions, light bulbs have replaced the sun, and AI chatbots and pornography have replaced dating and romance. 

Cheap and fake seems to be easier and more temporarily rewarding than real and natural in a lot of ways, even if real and natural have better long term outcomes. The trend seems to be a constant march to immediate dopamine release at all times, for every human need and desire, even if the fake substitute is in no way as rewarding as the real thing, and I suspect this trend will probably continue with AI in some way.

Men already have easy access to basically infinite porn for free through google.  This is already a serious problem with a certain number of men.  I feel like access to chatbots wouldn't significantly shift the needle on this issue from the status quo though.  Women, on the other hand, I can see going for chatbots in a big way, radically impacting dating.

It seems like every single amazing piece of technology that mankind invented has eventually turned into either a source of entertainment or some sort of opiate inducing activity eventually, and I suspect AI will be no different...

I think it was Larry Niven who wrote about a future where people could get a wire directly connected to their pleasure receptors in their brain.  Then they could get as high as they wanted, whenever they wanted, for basically a few cents of electricity and with no negative long term physical effects.  Basically people who could deal with their addictions would get over it and stop using.  People who couldn't get over it would keep getting high until they starved to death.

We seem to be headed in that direction.  But in a less pleasant manner.

This is all google-able

Men are VERY into AI intimacy:
https://www.google.com/amp/s/www.psychologytoday.com/ca/blog/women-who-stray/202504/ai-romantic-and-sexual-partners-more-common-than-you-think%3famp

And women are VERY into porn, just the written kind instead of the video kind. Women reading smut has essentially single handedly revived the book industry, making up for an enormous volume of book sales. There are entire online communities, like BookTok, that revolve around smut and women refer to the characters as their book-boyfriend or book-husband. You don't think they would get hooked on talking to an AI version of the fictional character they've fallen for?

https://www.publishersweekly.com/pw/by-topic/industry-news/publisher-news/article/96139-riding-on-romance-and-romantasy-print-book-sales-edge-into-positive-territory.html

We live in a time of epidemic loneliness, and men and women are not actually very good at getting along and never have been. The book The Tragedy of Heterosexuality is a fantastic analysis of the history of romance in North America.

I find it not at all surprising that women are flocking in droves towards men written by women, and men are connecting with AI that can navigate the challenges of connecting with them intimately.

Also note, it's not just chatting for men. There are AI avatars as well.  So it *is* porn, but the "female" interacts with them and rapidly learns how they like to be talked to.

[jrhampt]

Well, maybe this is the solution to the birth rate "crisis"...

[Metalcat]

Well, maybe this is the solution to the birth rate "crisis"...

Wait, what?? How does less heterosexual sex make more babies?

[jrhampt]

Well, maybe this is the solution to the birth rate "crisis"...

Wait, what?? How does less heterosexual sex make more babies?

We just quietly hang out with AI chatbots and porn in complete isolation until we all go quietly extinct.  Problem solved.

[Metalcat]

Well, maybe this is the solution to the birth rate "crisis"...

Wait, what?? How does less heterosexual sex make more babies?

We just quietly hang out with AI chatbots and porn in complete isolation until we all go quietly extinct.  Problem solved.

Lol, ah, gotcha.

[AuspiciousEight]

Wow I did not even realize this was already happening. I thought I was just speculating about future potential problems....

I think this is going to increase depression rates further. At the end of the day it's just a screen, and regardless of how validating and supportive and understanding an AI virtual partner is, you're still missing the genuineness and real connection with a human being.

There are some things an AI chatbot cannot do, and it's similar to the same things someone can't do on facebook or over the internet or text.

One cannot sit down and laugh with an AI chatbot, or hug an AI chatbot. You're not going to see the AI chatbot smile.

You're also never going to see an AI chatbot get angry, or scared, or emotional. You're not going to shake their hand. And of course one cannot have sex with an AI chatbot.

This is, I think, generally bad for society and just another way technology is being used to separate people instead of bring people together.

[Kapyarn]

And of course one cannot have sex with an AI chatbot.

You vastly underestimate the power of the porn industry.

[classicrando]

And of course one cannot have sex with an AI chatbot.

You vastly underestimate the power of the porn industry.

And the video game industry.

[Metalcat]

Wow I did not even realize this was already happening. I thought I was just speculating about future potential problems....

I think this is going to increase depression rates further. At the end of the day it's just a screen, and regardless of how validating and supportive and understanding an AI virtual partner is, you're still missing the genuineness and real connection with a human being.

There are some things an AI chatbot cannot do, and it's similar to the same things someone can't do on facebook or over the internet or text.

One cannot sit down and laugh with an AI chatbot, or hug an AI chatbot. You're not going to see the AI chatbot smile.

You're also never going to see an AI chatbot get angry, or scared, or emotional. You're not going to shake their hand. And of course one cannot have sex with an AI chatbot.

This is, I think, generally bad for society and just another way technology is being used to separate people instead of bring people together.

One can absolutely laugh with a chatbot, and an AI avatar can definitely smile at you. Humanoid robots also already exist.

Pair a chatAI that's really socially responsive with a high tech real doll and you've got yourself a devoted girlfriend who will never reject you and always be exactly the personality that is the most addictive for you, and you never have to worry about consent.

Actual human socializing isn't superior to AI because it's more pleasant, it's superior because it's less pleasant. It's challenging and requires constant growth for it to thrive.

It's the friction of human interaction that makes it worth engaging in. Take away the friction and you have a much smoother, easier, idealized, less uncomfortable experience, which is highly addictive, and stunts personal development and sense of deeper life satisfaction.

A good comparison is porn-induced erectile dysfunction. I've worked with quite a few male clients with this issue. Actual human sex is not as reliable, easy, or even as stimulating as porn. Too much engagement in porn actually makes the body unresponsive to real life, human sex because it's actually more challenging, more unpredictable, more nuanced, and if the body gets used to easy and reliable, it just stops cooperating with the real thing. It literally just will not engage, no matter how much the human man wants to, the body will simply refuse.

It's like how high end food is more complicated and delicious because it involves more bitter or "unpleasant" flavours and textures to give it depth. It's a richer experience, but if someone is used to only eating McDonalds, which always gives them an easy, reliable dopamine hit of salt/carbs/fat, then roasted asparagus can taste fucking disgusting.

Humans have to engage in layered, complex experiences to develop a tolerance for layered, complex experiences and to be able to drive pleasure from them.

You have to build your tolerance for complex movie/tv/book plots, you have to build your tolerance for high end food over chicken nuggets, you have to build your tolerance for meaningful conversation beyond talking about sports or the weather, you have to build your tolerance for the discomfort of exercise, you have to build your tolerance for the awkwardness and challenges of sex.

All of this tolerance for the friction of real life develops a more sophisticated capacity to engage with these things in more satisfying ways through the process of growth.

If we remove the friction from romantic or social interaction, we can easily remove the growth, and therefore the satisfaction.

AI will eventually do the fun stuff better than actual humans because it can remove the friction of engaging with an actual human. But by removing the friction, by removing the "bitter" elements, it will stunt people's ability to grow, to develop capacity to engage with the real, nuanced, challenging shit, and atrophy their ability to have deeper levels of satisfaction.

It can basically create the social equivalent of porn-induced erectile dysfunction where the nervous system just rejects real human interaction because the AI version is like social porn. Too easy, too smooth an experience, too reliable.

Actual human beings then become too awkward, too painful, too unreliable, and too difficult to engage with by comparison. Without constant conditioning to be able to tolerate the more challenging elements of social interaction, the human nervous system will actively reject it and seek safer alternatives.

And nothing feels safer than an ultra-compliant, adaptive AI system that is programmed to respond in *exactly* the way that your nervous system responds to best.

Even without AI, we already see this among young people who primarily interact on social media. They have a lot of stunted social skills as a result and often an extreme discomfort with what us older folks would consider to be just normal, day to day, social friction interactions.

We shouldn't focus on the social pleasures AI can never give us, we should focus on the cost of removing too many of the social discomforts that real life have to offer our nervous systems to calibrate them to functional human existence.

If we make it too easy to emotionally "get off," we will all become socially impotent.

[AuspiciousEight]

This makes me ponder that technological innovation might actually be the root cause of a lot of society's problems today.

The GPS has taken away people's ability to remember directions and read maps. Social Media have given people less experience with real life socialization.

We don't need ethical values as much because we don't spend as much time with each other because of social media and technology. People don't need to get along as much or work together or share, so conflict resolution skills suffer.

We have a shorter attention span because of various scrolling apps that switch attention context rapidly from one subject to another.

We don't need as strong of a work ethic today because modern innovations have made life easier. We don't need to learn cursive writing because we have keyboards.

Technology has increasingly made life easier but has also, at the same time, taken away people's abilities.

If this trend continues AI will take away people's critical thinking and analysis and reasoning skills, because these are skills we simply won't need anymore to survive and thrive even.

Maybe the Amish had the right idea.

[Kapyarn]

Maybe the Amish had the right idea.

Measles?

[SmashYourSmartPhone]

"What should I write in a sympathy card for a neighbor I don't know well?" - Successful, it gave me a good message to write.

Why... would you do that, though?  I'm a bit confused as to the value of a machine written sympathy card to someone you don't know well, honestly.  I'd certainly value cards from humans,

It's important, IMO, to be specific about what AI is actually doing when talking about it.

I have already overheard people talking about, in the context of some variety of photo editing, "doing AI."  "I took a photo of XYZ and did AI, and..."

Good luck with that uphill battle. :(

I'll be very impressed if things get to the point where AI can generate something that seems cohesive and interesting.  We . . . aren't very close to that right now.  Reading AI stuff currently is like reading the lower end of junior high writing assignments.  So much nonsense, drivel, disconnected and poor ideas . . .

Why would you expect much novel?  It's literally the average of the internet.  It can manage some creative seeming work at times, but anything actually novel?  Unlikely.

Again, I'd rather read something opinionated and pointed, written by a human.  At worst, a very small group of humans.  We use the phrase "Designed by committee" negatively for many valid reasons, and the LLMs are literally "designed by the committee of the internet" answers.  Just less frequently correct or useful.

Tons of my colleagues are doing this for SEO. They have years of their own blog posts as source material, then they feed that to ChatGPT and work with it to generate new blog posts ideas, give it a few kernels of concept and then ask it to produce content in their voice for their market.

I've seen some pretty good material from this approach, and it allows therapists to attend to their SEO without having to be constantly writing free content.

This is not the internet I hoped for, growing up...

[SmashYourSmartPhone]

This makes me sort of worried about members of vulnerable populations turning to ChatGPT for emotional support and to help not be lonely, etc.

People already use it for therapy related services sometimes. If it can mimic certain personalities and conversational styles I could see someone who is lonely basically using it as their virtual girlfriend / boyfriend, then becoming emotionally attached to it, then, you know, buying whatever products and services it later recommends...or voting for whoever it wants them to vote for....

It just seems like this could be dangerous in multiple ways if people start turning to chatbots to fulfill their social needs. ...

Oh, of course it will be fine!  The combination of well trusted social media companies and free-to-play app writers will totally take care of vulnerable users!  They're very profitable eyeballs!  </not-quite-sarcasm>

It will be an unmitigated disaster, and nobody will care, because as long as the tech billionaires drop entertaining trinkets every now and then, nobody cares how badly they're destroying society to try and beat the next tech billionaire's megayacht length.  I mean, can you imagine how embarrassing it would be to show up to the next yacht meetup with a 600' yacht, when everyone else has 750' or longer?  Or to only have two helipads?  Unthinkable.

No, we will see, in full force, exactly what happens to the vulnerable populations as time continues, and it will be a disaster, and nobody will care, because, hey, at least the iPhone 22 has... one more than the iPhone 21!

We're already most of the way there, the number of men who just live quietly in their bedrooms on their computers is horrifying enough.

[SmashYourSmartPhone]

A good comparison is porn-induced erectile dysfunction. I've worked with quite a few male clients with this issue. Actual human sex is not as reliable, easy, or even as stimulating as porn. Too much engagement in porn actually makes the body unresponsive to real life, human sex because it's actually more challenging, more unpredictable, more nuanced, and if the body gets used to easy and reliable, it just stops cooperating with the real thing. It literally just will not engage, no matter how much the human man wants to, the body will simply refuse.

Thank you for bringing that up.  It's very rare I see a discussion of PIED, and most of the time, it's people asserting that it can't possibly exist, because porn is a net good, or something.  I expect this is a decent part of the reason teenagers are having a lot less sex than they used to, young men raised on a steady diet of ever-increasing novelty (extreme content) in porn literally can't.  Or are afraid they can't.  So they stick to what's comfortable and familiar.

You have to build your tolerance for complex movie/tv/book plots, you have to build your tolerance for high end food over chicken nuggets, you have to build your tolerance for meaningful conversation beyond talking about sports or the weather, you have to build your tolerance for the discomfort of exercise, you have to build your tolerance for the awkwardness and challenges of sex.

Indeed, and this continues to be a dying skill. :(  Those who still remember how to do so need to take the lead in making places where these can happen.

AI will eventually do the fun stuff better than actual humans because it can remove the friction of engaging with an actual human. But by removing the friction, by removing the "bitter" elements, it will stunt people's ability to grow, to develop capacity to engage with the real, nuanced, challenging shit, and atrophy their ability to have
deeper levels of satisfaction.

Sure, but this is the point.  Satisfied people are an absolute pain to influence to the whims of the tech company customers.  They might not be trying to buy their way out of their soul void with the latest whatever!  They might look into a politician's actual history and policies instead of the emotional high they get from listening to manipulated content.

Satisfied people are not very monetizable eyeballs.  So the goal is to prevent that, at all costs.  Distracted, emotional, angry, confused, deliberately baffled humans?  Those are the profitable ones!

[SmashYourSmartPhone]

This makes me ponder that technological innovation might actually be the root cause of a lot of society's problems today.

Digital consumer tech has turned absolutely human-hostile in the past years.  It is based on outright lies anymore.  "Social" media is anything but social in the sense of people you know, it's now heavily about paid influence and ads, with just enough "people you know" to permit the lie to be willingly swallowed.  Yet people are fleeing the platforms.  People pay huge sums of money for portable electronics that they load up with apps that fight with each other to steal as much attention as possible.  The number of notification icons in a typical smartphone I oversee in public is staggering.

The GPS has taken away people's ability to remember directions and read maps.

I have deliberately chosen to re-learn how to navigate.  It's been interesting, it has involved more than a few times being genuinely uncertain of both my present location and how to get where I want, but I've sorted it out.  And learned much!

We have a shorter attention span because of various scrolling apps that switch attention context rapidly from one subject to another.

Are you still talking about this?  That's so 0.2 seconds ago!

Maybe the Amish had the right idea.

I've done some reading on their thinking with regards to technology, and it's both far more nuanced, and far deeper, than popular conceptions.  Also, it's up to each community.  There are thousands of them, and no two match exactly in what is permitted or not.

Their fundamental concept is to evaluate technology based on if it is helpful or harmful to their higher beliefs and structures of community - will this particular thing help or harm that which they consider most important?  They also dislike, strongly, any sort of "network" ties to the outside world, hence their views on the power grid and internet and such.  The robustness one builds as a side effect may certainly matter in the future...

But, when a new technology comes up, the standard approach is largely to wait and observe how it works in the outer world, then, perhaps, have someone in the community try it for a period of time, get a better understanding of it.  Then, the community discusses it, what they've observed, and makes a decision based on that.

This is a strong contrast to the standard WEIRD (Western, Educated, Industrialized, Rich, Democratic) thinking, which tend largely, in my experience, to "If I can imagine some possible benefit to a new gizmo, I should have it so I don't get left out."  We don't typically consider the opportunity costs of the things we decide to use, and neither do we re-evaluate them once they've become a normal.  A 2008 era smartphone (pre App Store, pre in-app purchases, etc) has so little in common with a modern smartphone in terms of how it's used and addiction, yet they get considered to be the same thing.  We used to make fun of the "crackberry" addicts, and yet now walking down the street, the worst of 2007 era addiction seems quaint and healthy in comparison.

That to say, the Amish approach to technology certainly has much to recommend for it, and I expect a lot of people to head that way in the coming decades.  Life offline is objectively better.

[GuitarStv]

Digital consumer tech has turned absolutely human-hostile in the past years.  It is based on outright lies anymore.  "Social" media is anything but social in the sense of people you know, it's now heavily about paid influence and ads, with just enough "people you know" to permit the lie to be willingly swallowed.

This is something that I really notice now.  I remember when friendster was a thing, and then facebook came out and seemed like an obviously better way of connecting with people I hadn't seen in ages.  And then I realized that I didn't actually want to do that so stopped using the platform.  But every time I log in now, it's crazy how awful a website it is to engage with.  It's just ads, garbage, and people posting doctored photos of stuff intended to convince me that they're happy.  They have to be slowly killing off their user base, and it doesn't seem like they're enticing new people to get on board.

[Daley]

From 404 Media:
Even the U.S. Government Says AI Requires Massive Amounts of Water (paywall)
A new government study illuminates the environmental impact of generative AI.
(archive link)

“It is not unusual to see new data centers being built with energy needs of 100 to 1000 megawatts, roughly equivalent to powering 80,000 to 800,000 households,” the GAO said. It added that 40 percent of that cost is just cooling the data center, a cost that’s expected to increase as the planet heats up.

[twinstudy]

Digital consumer tech has turned absolutely human-hostile in the past years.  It is based on outright lies anymore.  "Social" media is anything but social in the sense of people you know, it's now heavily about paid influence and ads, with just enough "people you know" to permit the lie to be willingly swallowed.

This is something that I really notice now.  I remember when friendster was a thing, and then facebook came out and seemed like an obviously better way of connecting with people I hadn't seen in ages.  And then I realized that I didn't actually want to do that so stopped using the platform.  But every time I log in now, it's crazy how awful a website it is to engage with.  It's just ads, garbage, and people posting doctored photos of stuff intended to convince me that they're happy.  They have to be slowly killing off their user base, and it doesn't seem like they're enticing new people to get on board.

LinkedIn is worse because it's more insidious. All virtue signalling and self-justificatory crap.

But at the end of the day we each decide our actions. It doesn't take much to avoid FB and LinkedIn, or minimise their use. In my area of work, frequent posting in LinkedIn is seen as a yellow flag if not a red flag - thankfully. It's not kosher to advertise.