We're Raising Kids to Prefer AI over People–and No One's Noticing
substack.com
substack.com
-- Sarah Connor - Terminator 2 (1991)
Ironic given how many HN commenters were tricked into upvoting this LLM junk article. Read the statement at the end - It was AI generated. It’s on a new substack. It’s posted by a new HN account. It’s LLM content mill junk from top to bottom and people were eating it right up.
Isn’t this just another way of saying that people are upvoting the headline without reading the article?
The headline was chosen as clickbait/ragebait because it feels interesting. The article revealed that that it wasn’t.
> This article was written in collaboration with AI. The tool helped shape the words—but the questions, the direction, and the discomfort it carries are human.
If I take their disclaimer at face value it could mean that the author used AI to fix their grammar. Does this make their arguments invalid? If the author wanted to just post LLM junk then wouldn’t they omit the disclaimer altogether?
I think it’s important to attribute the use of AI when submitting an article, but I’d be less inclined to do so if readers would label my writing entirely as LLM generated. Maybe the only solution is to not use LLMs and just get gud at English.
I can only imagine what that would say about the nature of dysfunctional childhood if society ever reached the point where the preference for personal liberation through AI arose.
Choosing a partner is not gambling.
As someone who has done a lot of research on child and adult development, I think AI as it exists now probably has the potential to do a better job than humans at many/most parenting tasks that it can do other than showing love (and that can go down some weird paths quickly with humans or AI).
Certain life skills, executive functioning, conflict resolution, curiosity, creativity, empathy — these are all things that I think AI (with the right prompts) can do better than most human parents.
Note that I’m referring to parents writ large rather than the almost certainly above-average quality of parents with which HNers have direct or indirect experience.
A parent interaction, game, friends, or the great outdoors, my child will reject 100 times of 100 if something on a tablet is available, which is why the time must be limited.
I've seen a twitter thread where a woman described conditioning her son to find screens boring at an early age by only showing slow train riding footage on screens around the house at all times when the kid was around, which is interesting. I wonder if that'd work well in general or just for that one specific child. It'd probably fall apart as soon as the child gets introduced to the fact that cool dopamine hacking content does exist on screens?
You could of course just use the tablet to show slow riding trains. At some point though you end up on an 8 hour car ride or flight, and without an army of novel toys it's just far more practical to hand the tablet than have the child screaming bloody murder the whole time because they are bored. This is one of those things that makes the parent sound bad but parenting is 24/7 365 days a year and sometimes people need a low-effort break for an hour, even the child.
I'm very sure it is harder to achieve this now than then, I having been born at a time when color LCDs were just beginning to make tentative steps from laboratory prototype to consumer product. Unless Hoel is taken for the liar I've seen no cause to think him, though, it surely still can be done.
Easy for me to say, though, never having had or raised a child. I like to hope I still could recommend the method if I had, but who knows? I do think it asks uncommon interest from the tutor.
Consider for example the process of casually learning about a subject in a non-academic scenario. People frequently will search online, ask on forums/chat and watch videos. Search results are widely critiqued for being poor these days, so people often turn to asking communities for specific questions.
Sometimes the only communities available are beginner hostile, expecting some prerequisite understanding and even being antagonistic toward knowledge gaps (mostly because it's uninteresting to them and they've seen so many of such questions it introduces a jadedness).
It's scenarios like that where non-judgmental, always available LLMs stand out in contrast most. I have wondered though how that might shape the dependency on LLMs broadly if people get used to not having to deal with the rough edges of humans.
If anything, the existence of an Internet supporting direct discussion like this between strangers like we should make vastly more nuance available at the median now than then.
What a shame we all seem to spend so much time making such poor use of it! I grew up on the dream that global communication would lead to global understanding. Imagine my disillusionment on discovering at last the besetting, lethal flaw in this beautiful revel: that we've nothing better than humans to give the job of doing it.
I don't think LLMs are going to remain at perfectly timed emotional responses, calm voice that never judges, and answers that always fit. We are already seeing them adapt more, through user personalization, memories and access to past conversations. So LLMs move from static to recursive dynamics.
But what happens when billions of people create interactive logs with AI? What is in those logs?
- students learning
- problem solving
- planning tasks
- search and research
- psychological support and therapy
- companionship
So it's no longer an answer machine. It's a problem solver, a teacher and a generative-teleological agent. It help us clarify what we want, understand ourselves better. Not just solving our tasks, now it helps us choose what to do next.
Should we fear its effect or not?
It can be the greatest teacher, with a vast pedagogical experience. It can be the best problem solver not because it is smarter than us, but because it is learning from everyone's mistakes. It might be able to help more people in need than psychologists, always on, never bored, as the OP said.
> (there can be a whole different discussion what better means)
This is the major road block to any fruitful discussion of trade-offs. Two people, regardless of intelligence or thought process, will often have diverging definitions and goals.
As for the kids, I think it doesn't matter whether it is a TV, a tablet or an AI something. It's simply either an educational/developing interaction or not. You just need to have control over it and have limits for things like this. It's the same challenge, but simply in a new form. Kids love doing interesting things, so parents just need to stop being lazy and find good ways for their children to spend time.
I think a lot of it has to do with the attention economy. Our attentions are so occupied by social media and we are so accustomed to its rapid feedback and endless stream of novelty that we get frustrated and annoyed by the slow speed of communication with other people. So often I find my attention wandering while someone slowly explains something to me and they will often notice and be offended or hurt by it. These interactions make relationships difficult to maintain and I don’t really know what to do about it!
If you place that ability to communicate with confidence and authority in an individual with male gender traits who is tall, fit, socially attractive, and has a deep voice - nearly everyone around them will do what they say; even to the extent that doing so many be against the best interest of the individuals listening.
AI speaks with utmost confidence in every interaction - not just kids, but all humans will have a preference for it.
The Rationalist and Dark Enlightenment movements are just the start of what will become an ever-increasing wave of people claiming to have solved all of society's problems from first principles, and we're going to have to spend ever more energy explaining that reality has a surprising amount of detail.
The question is whether this can be taught without turning kids into cynics.
Also: This article was written in collaboration with AI. The tool helped shape the words—but the questions, the direction, and the discomfort it carries are human. Oh the irony.
Not sure I agree with the post, but the effects of growing up with ubiquitous AI on our children will be profound and worth pondering and discussion. Just like the effect of growing up with smartphones has been profound (and, as a father of such children, the effect is on balance more detrimental than useful).
> – a perfectly timed emotional response
> – a calm voice that never judges
> – an answer that always fits what they needed to hear
And what’s bad about this? Maybe humans need to rethink their behavior if machine is better?
Whole article is just authors screaming insecurity.
So ... that makes the machine better?
I believe the term for a human that acts like that is:
> if machine is better?
Define "better" ? What are we measuring ? Instant self gratification ? Long term emotional independence ?
What's bad about this is when these children would need to fit inside a certain circle with other people, who don't behave like machines do. Circle like school, work, or family. These children might have issues accommodating there.
You may then ask, "maybe humans don't need schools, work, or families," but that would be a different conversation.
> We’re not raising emotionally intelligent kids. We’re raising kids to navigate human unpredictability as if it’s a design flaw. Because when you grow up with a machine that always gets you, messy human behavior feels broken. We’re not preparing kids to handle people.
I don’t think there’s anything wrong with escaping into fantasy in the right time and place, but young kids (and even well-adjusted adults) can have problems self-moderating and letting fantasy substitute for engaging with reality.
We need rough edges, we need some level of inconsistency.
If a child is grown up on machine, they’ll prefer machine for friends, dating, colleagues.
We’re already seeing a subset of the population who are less physically social turn to AI to fill the gap. Not necessarily a bad thing for adults, but preferring machine over humans in place of friends during a childs most formidable years is a recipe for societal disaster.
What is bad about this is that we exist in a real world, with other imperfect humans that we need to learn to interact with, and sometimes very tough social situations we will have to learn to navigate
Children who are overly coddled and never challenged grow up to become insecure, entitled adults. They expect everything to continue coddling them forever
> A calm voice that never judges
If you are never judged, you never improve
I'll boil this down to the simplest possible explanation of why your statement is idiotic -- children who are trained to prefer AI, will never learn how to form friendships with their peers (aka other children, who won't always give perfectly timed emotional responses or always be calm or always answer what they need to hear). Other children are not able to "rethink their behavior" yet because they are children still.
A world in which, I ask my child "do you want to play with Timmy next door, or stay inside and play with Alexa/Siri/etc." and my child always prefers to pick Alexa, is one of the most dystopian outcomes I could possibly imagine for childrearing.
Forming friendships and human connections is a skill. Learning new skills is hard and not always fun. A soothing AI companion that always says what you want to hear, is going to trap children in a dopamine loop that prevents that kind of social skill development from ever happening.
If you believe the argument makes enough sense to justify replacing adults, then why do you think children will still want to play with each other, instead of just playing with AI? If anything, other children are MORE likely to display shitty behaviors than adults, given that you know, they're fucking children who don't know any better.
Have you spent any time around children?
I’ve spent years being one.
It is precisely because I go outside why I consider AI superior teacher.
I have yet to see a single AI proponent commit to giving their kid over to AI, if they even have one. SV CEOs are sending their kids to hardcore techfree schools and limiting their own kids access to their tech, all while insisting AI and their tech poses no risk at all to kids and that any reservations are "fear mongering."
It shows. The whole "So What Can We Do?" section lacks substance.
afaict the substack account was created 17 minutes ago and this is their first post. The HN account that posted it was created 49 minutes ago.
People are manipulative, combative, dumb, unethical. I prefer AI over the average person, but I still prefer really exceptional individuals over any AI.
The most dangerous animal on this planet by far is the one that doesn't think of itself so. Even still, I would rather face an armed and angry human than an LLM with a goal best served by ending my life. I can smell how likely the human is to kill me in the next thirty seconds. The machine, not so much, nor will it face even the briefest moment of emotional difficulty at the prospect.
It isn't that I don't get what you're saying, goodness knows. Sharing this species with a planet is no bed of roses. Still, I've spent decades learning to understand the animal. No one has spent nearly that long learning to understand the machine.
I believe you may be also overindexing on the circumstance in which attachment fails from the parental side. Again, if so I can empathize, but wish to note other outcomes do occur.
Though those who know me would take it as an obvious joke to hear me say, I don't actually care that much for most humans, because the median human by my standard barely qualifies as sapient; they don't deserve to be as miserable as they insist upon, but at the same time they do make it hard to feel really safe. Still, one hopes one has the sense not to assume anything that isn't human must by that sole virtue necessarily be trustworthy. Especially if humans "invented" it.
I feel like I missed the boat, this sounds like a fantastic self-soothing option.
Of course the downsides are just as myriad as the upsides. People used to care how much Google knows about them, now they willingly tell their private thoughts to ChatGPT or Gemini. Also AI is great at mimicking compassion, but it doesn't actually feel anything, it's closer to a calculating psychopath mimicking emotions (without even going into the whole debate whether current AI can reason or just does something that to us looks similar to reasoning). And of course most importantly, everyone "connecting" with AI instead of real humans is pretty dystopian on its own, even if we ignore all the ways this can be abused by bad actors
The ai will notice, and the corporations that run them.
> This article was written in collaboration with AI.
If you didn’t catch on, this is a newly created HN account posting a link to a newly created Substack which has newly created LLM slop. It’s ironic that the content is about children getting immediate satisfaction from LLMs telling them what they want to hear, because that’s what’s happening in this article.
I suspected it was AI generated because it followed the pattern I see from LLM-promoted debate arguments: It puts forward a conjecture, treats it as fact, provides no evidence or support, and then circles around the conjecture for paragraph after paragraph in hopes that the reader will accept it as true.
This content is effective for hooking people who already believe something to be true because it mirrors the result they want to see. Do any critical thinking on the article and you see that the arguments are effectively circular: They only make sense if you presume the premise to be true from the start. It’s as though the LLM has become good at taking a belief and doing a style transfer to rewrite it in the form of an argument, but it can’t actually fill in the parts of the argument.
The suggestion in the middle that we start deliberately training AI to be imperfect should have also been a giveaway that this article wasn’t serious
I agree that this article is slop, but where can I see the rest of their substack articles? https://askswithoutpermission.substack.com/profile/posts shows no posts at all so maybe it's just broken.
As was the HN account that posted it. Note the green username.
This is what it looks like when someone decides to try fishing for traction with AI junk. They probably went from idea to having the blog up and posted in 30 minutes or less. They’re probably working on a different iteration which will get posted here on HN again soon.