Live data from Hacker News

Building a real-time AI tutor for 5-year-olds

ello.com

251–260 of 460 posts

Re: Building a real-time AI tutor for 5-year-olds

#251

Earlier quoted context omitted.

why would a 5-year old need a tutor?

because of bad grades?

At 5 years old, in the US and UK at least, they're either in preschool/kindergarten or not in school at all yet. Grades aren't a thing at that level and I have to imagine the same is true in other areas of the world.

Re: Building a real-time AI tutor for 5-year-olds

#252

Earlier quoted context omitted.

> In practice, yes. In theory a machines can do about everything a human can, but better and cheaper. If you could time-travel back to your 5y old self, would you prefer to be taught by AI tutor given the current state o/t art, or taught by whatever teachers you did have when you were 5? (with all the existing hallucination, breaking through guardrails etc problems of current AI in mind) If you'd have a ~5y old yours…

> If you could time-travel back to your 5y old self, would you prefer to be taught by AI tutor given the current state o/t art, or taught by whatever teachers you did have when you were 5? (with all the existing hallucination, breaking through guardrails etc problems of current AI in mind) I had to choose between my elementary school teachers and something like Claude Fable 5 with a good teaching-focused harness, I w…

Your average 5 year old will also gleefully attempt to eat an entire gallon of ice cream in one sitting so make of that what you will.

Re: Building a real-time AI tutor for 5-year-olds

#253

I think teaching a child to trust an LLM from a formative age is horrifically irresponsible. If anything, an app should be made where a child learns to correct an LLM's mistakes and learn that it isn't trustworthy. Actually, better, don't put an LLM in front of children. At all. EDIT: If a use case is for children who can't afford good education, then use an LLM to make educational materials for children, review them…

Perhaps human teachers would model correct grammar?

vs "you like dinosaurs! me too!” From the blog post.

Re: Building a real-time AI tutor for 5-year-olds

#254

I think teaching a child to trust an LLM from a formative age is horrifically irresponsible. If anything, an app should be made where a child learns to correct an LLM's mistakes and learn that it isn't trustworthy. Actually, better, don't put an LLM in front of children. At all. EDIT: If a use case is for children who can't afford good education, then use an LLM to make educational materials for children, review them…

Perhaps human teachers would model correct grammar? vs "you like dinosaurs! me too!” From the blog post.

Perfect grammar is less important than comprehensible semantics for all but the most niche linguistic applications.

Re: Building a real-time AI tutor for 5-year-olds

#255

I'm torn about this. I primarily think that a a kid who won't pick books is a failure of the family of not noticing their interests. When i noticed my oldest 3 yr old was obsessed about cars, i bought him a car encyclopedia. He probably could not read most words or did not understand them. But pretty soon he was telling me about car models that i did not even recognize myself. And as i saw interests, i kept feeding t…

Children are going to witness or experience violence at some point or ask questions about it. They will probably see some at school. I think it depends when and how you introduce it. For example, there is a heavy focus on refugees in western schools and media, but at some point a child is eventually going to have to ask what refugees are running from. (Torture and violence.) It's hard to avoid discussing violence in…

You’re not claiming that most refugees are running from torture are you?

Most are economic refugees running TOWARDS a better economic outcome than is available in their home country.

Re: Building a real-time AI tutor for 5-year-olds

#256
As someone who has been engaged in children's lives in the age range of 4-9 that have struggled with school, I can't imagine using AI being a net benefit to them. You know what helped me succeed in helping these children more than anything else? Listening to them. That's it. Listening to them to understand their viewpoint and perspective on the world, their interest areas, and the things they were curious about. Then you incorporate those things in how you teach them through direct one on one instruction. It's that simple. My niece, as an example, is now at the top of her class in most subjects and has realized she has a passion for history and geology.

Not every child has the benefit of involved adults in their life, but I'd rather solve that problem than think AI is going to fix this. You're proposing taking one of the most vulnerable populations in the country and manipulating them (intentionally or not) to be reliant on one of the most exploitative technologies that has ever existed. I don't think your intent is evil, but the product here is evil. Remember that the path to hell is paved with good intentions.

Re: Building a real-time AI tutor for 5-year-olds

#258

Earlier quoted context omitted.

> Ask them how many r's are in strawberry. 2024 called and wants its talking points back. > Teachers don't make confident mistakes because the student asked the wrong way. Teachers can be held accountable. Teachers can learn. Teachers can love their students. All of these can be true, and at the same it can be true that a child tutored by a good quality state-of-the-art LLM with a good teaching-focused harness could…

> 2024 called and wants its talking points back. It's a classic example of the kinds of questions LLMs get wrong. There's plenty of others. Not sure your point here. We can easily find things a 4-9 year old will talk about that an LLM will get wrong, hallucinate, etc. > a child tutored by a good quality state-of-the-art LLM with a good teaching-focused harness could have better learning outcomes than a child without…

> It's a classic example of the kinds of questions LLMs get wrong. There's plenty of others. Not sure your point here. We can easily find things a 4-9 year old will talk about that an LLM will get wrong, hallucinate, etc.

I'm not participating in the debate, but fyi I have just asked haiku, sonnet, opus and fable that question and they all got it right. Tried with the letter s in portuguese word "sussuarana" (mountain lion) too - same thing.

Re: Building a real-time AI tutor for 5-year-olds

#260
post #226

I'm torn about this. I primarily think that a a kid who won't pick books is a failure of the family of not noticing their interests. When i noticed my oldest 3 yr old was obsessed about cars, i bought him a car encyclopedia. He probably could not read most words or did not understand them. But pretty soon he was telling me about car models that i did not even recognize myself. And as i saw interests, i kept feeding t…

> At some point, you need to watch out for cynical/nasty series. In fact, all of the books we purchase are ranked against peers they have read in the past, or those we know to actively avoid buying due to cynicism, sarcasm, or open disdain towards adults (Wimpy Kid). yeah how do you "know" this? our kid is obsessed with wimpy kid and a few other similar series and "cynicism" nails it perfectly, is there a list? of wh…

Funnily enough, I was using ChatGPT for the same purpose a week or so ago. You can ask something like:

> I'd like recommendations for book series that avoid self-pity, misanthropy, nihilism, and over-indulgence in self-identity analysis. Lessons on character building and virtue can be present but ideally without overt moralizing. Characters should simply inspire through their actions. Starting from Curious George, and working toward things like LotR. Background themes should lay the groundwork for stoic and transcendental philosophy later.

If you then ask for series to avoid, at least in my session Wimpy Kid is explicitly named. Or more succinctly it basically told me I want pre-1970s children's literature and adventure fiction and stay away from contemporary school-market YA.

Post reply on HN