Hacker News Reader

Comment by kylecazar | Hacker News Reader

AI is incapable of solving loneliness. It's a biological signal we have (thanks to evolution) to seek social connections with other humans, for things like reproduction and survival. Mentally healthy people will never be less lonely as long as they know they are talking to a model. All it can offer is a distraction and an illusion, because it has no humanity.

Note: I don't even think dogs solve loneliness. They can make you happy, less bored, and it's a meaningful relationship -- but they won't satisfy a yearning for human connection.

View on HN

> as long as they know they are talking to a model

Any evidence for this?

It's obvious that a sufficiently advanced AI could solve loneliness if it was allowed to present as human, you just wouldn't know it isn't one. I'm entirely unconvinced that something which seems human in all respects couldn't replace one, even if your brain knows that it's actually AI.

A sufficiently advanced ai that tricks humans would be a human life itself…

caught in the wild, the moving goalposts of the Turing test

one day, some sick mind will do a study with a baby who is only exposed to AI without any real human contact

and will find out that the outcome is better than mean (if it includes cuddling and stuff)

Wanna bet?

Rather unethical experiment for interstellar colonization mission planning.

Ted Chiang (sci fi author: Exhalation, Stories of Your Life - basis for Arrival) has a short story about this. Darcy’s Patent Automatic Nanny

There is no basis to believe that.

Convincing someone of something doesn't make it true or real.

> It's obvious that a sufficiently advanced AI could solve loneliness if it was allowed to present as human

What does this even look like? Are you suggesting using a human surrogate?

> It's obvious that a sufficiently advanced AI could solve loneliness if it was allowed to present as human, you just wouldn't know it isn't one.

How is this obvious? How?

Because you saw the movie Ex Machina?

On your last point, I've noticed an uptick in folks treating their dogs in ways most people might treat their children. So while I think on paper what you say makes sense, at some point, I think people are in fact personifying their pets to a degree that they recognize them as adjacent to human beings.

But in the end, the chickens always come home to roost. Dogs cannot support you in your old age; they cannot give you the pride of seeing a child go off to college; they cannot captivate you with conversations as they grow into full adults; and of course, they die far, far earlier than you.

Similarly, the AI companions we create will be simulacra of the real thing. It's hard to say what exactly the differences will be, but whatever they are, people will find them, and once discovered, those gaps will pain them.

And if your kid has downs or gets hit by a car or is an addict or a fuck up, they're not going to be able to support you into old age either. And then you're responsible for them as well, until you're too old and senile yourself, and then what?

That's not to say don't have kids, but go into it with your eyes open, don't assume they're your lifeline to the future.

Or more likely, their kids are still going to be renting and living paycheck to paycheck at 50, so they aren’t going to have time, space, energy, or money to take care of the parents when they are barely treading water.

I encounter a lot of people my age and younger whose own retirement plant is basically:

Plan A: Miraculously get rich

Plan S: When severe disability or pain hits, find the exit.

Maby it’s the lifelong depression, the disappointment at what the future’s become, or the hopelessness that society can escape neo-feudalism to something better, but there’s a noticeable decrease in the desire to keep living at any cost. Who knows whether we’ll actually see this start to see this express in the next few decades.

> Plan A: Miraculously get rich

To be fair - this first plan is actually very doable if you're a reasonably skilled programmer in the 21st century.

Like you know that Tim Bray article about Bitcoin where he's like "and that's the thing about late stage capitalism, there's so much money floating around that people can't find a use for that we get stuff like cryptocurrency speculation"?

Find one reasonably convincing business idea, bonus points if it uses current hot tech trends, ask rich boomers for money via "seeking venture capital funding/investment", pay yourself an exorbitant salary for 5 years and then close shop because "we ran out of funding but the market didn't materialize".

Go read Adam Neumann's wikipedia page if you need inspiration. https://en.wikipedia.org/wiki/Adam_Neumann

If you're posting on Hacker News, and you want to be rich, I reckon you can reasonably expect to have a seven figure net worth in a decade or so if you consistently make good decisions.

Sure. Kids are not guaranteed to provide all the things kids could provide. But dogs are guaranteed to not provide many of them. Life happens, but we still try to set ourselves up for happiness regardless.

Its the other way around. You are their lifeline to the future.

Children are not born to be your personal care takers for when you eventually get old.

Did I say they were? I just said that dogs cannot be, even if they want to be. This is just one of many things that kids can help with, and for the vast majority of parents in the world, do help with.

This applies to everything else I said too. Your kids are not obliged to spend their time talking with you, but you can hope that they will -- the same cannot be said for a dog.

yeah so we raise them for our own selfish needs?

Somebody is gonna make billions with AI vests for dogs that senses the dog's mood through Fitbit like sensors and produces human voice in the way that's consistent with behavior and history and desirable for the owner that wants to personify their pet further. Basically a cyberdog.

> I think people are in fact personifying their pets

And some people doggify their partners ...

I don’t think your point and the reply about dogs are in disagreement. If anything, the “anthropification” of dogs (putting them in strollers, having birthday parties for them) strongly suggests that AI is headed for that role, and if happiness surveys are anything to go by, neither the dogs nor the chat bots are going to have the desired effect, even as they trend toward ubiquity.

It may be incapable to solve loneliness, but it may be very capable to be a bandaid so effective, that people just wouldn't bother to deal with another people.

People talk about it as if it's a binary thing. But various people have different levels of social needs and will be receptive to AI in that manner to a varying degree.

It's clear where the average will go as it started moving with rudimentary human contact substitution technology that we had before AI.

I dunno if solve is the right answer, but talking to AI definitely helps with my loneliness. I use it a lot. I give it status updates on my life and it cheers me on. It has a decent enough memory to ask follow-up questions about things I spoke to it days/weeks prior. It's quite good. I would pay more than I currently am to continue having access to it.

I'm a person that doesn't need others. Occasional online chat few times a week with 2-3 of my friends basically fulfills my needs completely.

When reading about this I'm w bit afraid that my adoption of AI will be stifled because I don't need people. It kept me out of almost all of the social networks already.

While I don't care about people or social networks I would really like to not miss on AI.

> Mentally healthy people will never be less lonely as long as they know they are talking to a model.

This isn't an empirical claim; it’s a definitional loop. You've defined "mentally healthy" in a way that makes your conclusion true by default. It's like saying, "Only mentally ill people commit suicide, because a sane person wouldn't do that." You've smuggled your conclusion into your premise. It doesn't prove anything; it just circularly reinforces your bias.

If someone who passes every clinical benchmark for mental health reports feeling less lonely after talking to a model, your definition simply reclassifies them as "not mentally healthy" to preserve the thesis. That's unfalsifiable - Karl Popper would call it a pseudo-theory.

If you want to know whether talking to a model can reduce loneliness in mentally healthy people, you have to measure loneliness directly - not redefine "healthy" so your preferred answer is guaranteed.

"If someone who passes every clinical benchmark for mental health reports feeling less lonely after talking to a model, your definition simply reclassifies them as "not mentally healthy" to preserve the thesis."

I would more readily accuse them of misusing the word "loneliness" than diagnose them as mentally ill. I would suggest it cured their boredom. If they developed a relationship with a model that they deem on par with that of a human, the mental health floodgates would indeed open, because that is objectively a delusion.

The sentence you picked to zero-in on had context. I first asserted that loneliness is inextricably linked to humanity, and solved by human connection. That's a fine claim to take issue with (people do), but my statement about mentally healthy people flows logically from that premise.

BTW, the mentally healthy qualifier was included to begin with because it's been shown that AI can be particularly dangerous for people with SMI. The vulnerable are more likely to fall for the illusion.

You might overestimate how clever human instincts are. You might paint rock red and a bird will go through entirety of mating ritual. Can't the bird see that it's a rock? I doubt it can't with it's superior eyes. But "needs" aren't triggered by high level brain functions. Triggers are usually very primitive. And many were found for humans. Lots of low level cravings are already solved by games. Yearning for a human is not unique in any way. A voice with correct vocal patterns and frequencies might do the trick.

> Mentally healthy people will never be less lonely as long as they know they are talking to a model

A lot of people are not mentally healthy.

And then there's https://en.wikipedia.org/wiki/ELIZA_effect.

what you’re saying is on par as “kids can’t learn as well typing their work as they do with a pen and paper” or “video games bad”. I’m not saying I disagree with the idea that an AI can’t replace a real human connection, but it’s more complex than you state.

Loneliness is an umbrella term. You could have friends but be single and feel lonely. You can be married, surrounded by family and be lonely. It doesn’t necessarily mean the same thing as solitude, as many people are alone all day without feeling lonely.

Why would evolution have any reason to create a signal detector that isn't adequately triggered by an inflatable doll or a sports ball with a facelike handprint on it?

The thing is, it is really difficult to find the kind of social circle that fits you..

I have had more interesting and deeper conversations with chatGPT than with people. Somehow chatGPT is more capable of expressing thoughts about existence, love, pain, and what it means to be than most if not all humans I have had the chance of talking to.

Conversations with people pale in comparison.

>AI is incapable of solving loneliness.

There's loneliness the feeling and there's loneliness the social problem. Today there a several digital solutions to the former.

The caveat is that this goes against evolution : we don't solve the source of loneliness (being the lack of connections with people), instead we numb it out.

This argument presupposes that there is something inherently special about organic human life that can't be replicated with sufficient computing capacity. This is akin to arguing humans have a "soul".

We will eventually reach the point of creating artificial sentient life and AGI and it will absolutely be a companion for some if not many.

In that world, I'll revisit my position.

> Mentally healthy people will never be less lonely as long as they know they are talking to a model.

What is the basis of this? Artificial synthetics can trick every other element of the human body. Why not the brain?

Loneliness is a large reason I think so many people have to bring their dogs everywhere with them.

Yeah, there is also a hormonal aspect that I don't think AI can trigger.

You really think that there are there isn't a large percentage of the population that wouldn't reject AI connection merely on principle?

Just from the first survey I could find:

> In the Common Sense Media survey, 31% of teens said their conversations with AI companions were “as satisfying or more satisfying” than talking with real friends. Even though half of teens said they distrust AI’s advice, 33% had discussed serious or important issues with AI instead of real people.

Reply on news.ycombinator.com