Thus far, 7yr and 10yr old nephews have not been introduced to LLMs. These kids already write code, and not introducing them to LLMs is somewhat analogous to the fact that they have not been given calculators for math.
At what age/point is teaching them about the availability of LLMs a good thing[0]? They will find out eventually, is a parental introduction better than just a random introduction?
[0] The creators of Django, Redis, and Linux believe that LLMs are not useless, so let's agree to move beyond that for the sake of argument.
The age of reason is 7-12, and varies, but generally prior to which they cannot biologically detect deception perceptually.
Given observed behavior of LLMs, you must think of LLMs as the most pathological type of liar. You don't ever give those types access to your kids when you are actually fulfill parental roles.
Maybe they should learn to read, write, do basic knowledge research first before outsourcing everything to a machine.
And yes, to go with your analogy of "not introducing them to LLMs is somewhat analogous to the fact that they have not been given calculators for math". They shouldn't have calculators before they can multiply 5x7 themselves.
We can't outsource very basic knowledge to machines if we want to stay lord over the machines.
When they're in their late teens is still early enough to be lazy and have LLMs do the work for them
I suspect households motivated by functional concerns will be far less likely, if at all, to introduce any access to AI and households motivated by political concerns will introduce access to AI at any age but only if they agree with the content or bias of the AI.
We've let our kids play with LLMs by having conversations in voice mode and generating images. The youngest one likes doing this, but it's a novelty, not something that he does all the time.
For academic work, we've had success using Perplexity (with parental guidance) for the older kids' projects that require Internet research. The ability to get an overview of a topic at a moderate level of complexity with links to other sources is beneficial. This isn't a substitute for doing in-depth research in the library or with actual peer-reviewed articles, but they're not yet at that level of depth.
At the college level, the most important lesson we're trying to teach is using LLMs as a source of ideas, suggestions, and feedback to advance your work, rather than as a tool for generating finished work. I often phrase this as "collaborating vs. delegating". I want students to think critically about their ideas and repeatedly iterate with LLMs in the loop to help solve the creative problems they encounter - but without outsourcing their own vision for the project.
My colleagues are seeing good results across multiple disciplines using LLMs for topic development and pre-writing, so I'd encourage leaning into that role, as opposed to jumping straight into text generation.
We've also learned that students benefit from a clear process with specific example prompts. Using AI well requires developing critical thinking and self-reflective skills, so there's a process of maturing that comes with time and exposure.
If you're interested, here's an example research assignment I've used in my own classes with some specific prompts and suggestions for different phases of the writing process:
Reasoning and problem solving is a CRAFT, and if it's not practiced it will atrophy.
Let's not fall into the trap of "LLMs are just a tool like a calculator". An LLM is like a calculator where you can just say, "Here's a math problem please solve it for me."
I'm also going to add some homeschool family perspective, as a father with several children. Hopefully this is helpful for other parents or soon-to-be parents.
Although there are basics, there truly is not a one-size fits all, even in a single family. From the start, our 2nd had a drastically different personality than our 1st, for example. Doesn't mean that you parent differently necessarily, but it's always a good idea to allow for an open perspective, especially when talking with other parents. Not only are people touchy about how they parent, there are simply many variables when it comes to being intentional about your family. It's okay to have different ways of doing it.
That said, as a traditionally-minded homeschooled household, I'll share what's worked for us.
We have only introduced technology in a limited, time-based, supervised capacity. The focus is on using it as an educational or research tool. Our children are not even aware social media or Youtube really exists. So, for example, we have done small tech projects to understand how things work (Arduino weather station, etc) or also, they're allowed to look online at pictures of dog breeds to their heart's content.
Having a background in linguistics and tech, with LLMs, I'm far more hesitant. Like with social media, I do wonder if we really understand yet how it might affect young minds. Of course, there are also the more Orwellian and political/ethical dynamics involved too.
Instead, we've focused heavily on reading and critical thinking skills, as well as them having a chance to have real conversations with adults. They regularly check out 50 books at a time from the library. Our children will likely graduate college at 16yo. They love learning and can have engaging conversations on difficult topics...with people of every age.
You might hear it in my voice, but it's hard not to be proud of this, especially in this day and age.
Anyway, two other tech questions simply to provoke thought...
- When did you first enroll your children in facial recognition profiling and tracking?
- When did you introduce firearms education to them?
Some simple system prompting on a local LLM can help you get there on your own believe it or not. A custom GPT on ChatGPT is another option. This will take an active involvement from you, because it’s a pretty much figure it out as you go for all of us.
Augmenting a parent so their child has a world class tutor is incredible stuff.
Both for "when do you introduce in a supervised setup" and "when do you let them use in an unsupervised setup".
LLMs are new to us, but for them it's not more new than the rest of the internet. It can be harmful, but probably less than social media.
LLMs are a tool like a calculator or Wikipedia. Kids will use them whether you like it or not. It’s important to guide and mentor them to use them productively.
That said, I think generally 6-7th grade is where that starts happening. Maybe in the context of coding something like CodeLlama is a good enhancement.
kidding aside i think its good to explain it in laymans terms as they grow. eventually they will reach college and have their papers rejected by an anti plagiarism ai and it will piss em off lol.
The upside is that before ChatGPT he showed no interest in written communication at all and his school was genuinely concerned. Even an educational psychologist couldn’t properly assess whether he had an actual learning disability over and above his neurological conditions because instead of taking the psychometric tests he ate the papers. Just to be clear, he’s not intellectually disabled in anyway, he’s just so defiant that he blatantly refused to engage in any psychometric tests, and he stated that his IQ was his private business. Instead of being annoyed at all his letter writing, the school have taken his letters as an indication that he has some ability to read, write and comprehend, as all the letters he has written have been adapted by him to suit certain nuanced circumstances that would either require detailed prompts, or alternatively heavy editing. I have to say the school’s response to the letters has been refreshing. They wanted to see what else he might use AI to come up with, or how else he might use AI to engage in learning. His school has been allowing him to use AI to complete non-assessed school work. Everyone is so happy because he firstly started doing school work, but then, he’s also started doing some school work, without using AI now as well. As a parent, I think if that was the aid he needed to build his self-esteem enough for him to feel able to engage in learning then I’m all for it.