Your policy prediction is spot on. The tech majors will pretend no kids use their platforms, and the kids will pretend they're older, and then there will be a panic
Arvind, thank you so much for sharing this. Here is an op-ed my father wrote when I was five about the same topic: https://bit.ly/MacChildPlay92
1. I appreciate your defense of the computers in the upbringing of children, and I think you are right about saying that "screentime" is not a category. I believe a healthy relationship to technology involves some long stretches of "quiet time" — I think Google broke us a bit in two ways: Because we can Google anything, 1) we think all questions are worth asking and 2) we sometimes don't think about questions on our own enough.
I think this is great, especially if in the future the parents have more customisability (rather than it being in the hands of OpenAI etc.). In that case, parents might come to trust the AI more than, for example, teachers... as the AI they can programme to be completely inline with their principles rather than those of the ideological teachers (I.E "Come on little Jimmy, don't make a sword from Lego, that's the patriarchy.")
I'd much prefer my kids were with AI whose ideological grounding I programmed.
"So far, the bot has not output any harmful content. There have been a couple of borderline cases. One of the stories involved a character dying of old age. Death is not an off-limits topic in our family, so this was not a problem for us."
This does suggest there might be some cases where the bot doesn't say anything *inappropriate* but still might introduce a new idea that parents would want to discuss in-person. I wouldn't want ChatGPT to be the first to tell my toddler that the family dog isn't going to live very long. Religious parents might be especially cautious. But it seems that an LLM whose training data is specially tailored toward children could be an excellent edutainment toy. (Just make sure kindergartners don't ask it for help with math problems...)
I also wanted to say that I really appreciated your recent tweets about using GPT-4 for looking up papers! (I don't have Twitter but read a few accounts.) GPT-4 seems as bad at *reasoning* as GPT-3.5, but it didn't occur to me that it would have a much stronger grasp of its own knowledge base for that kind of search. There are still a lot of hallucinations to wade through, but I finally found a paper about classical conditioning in nematodes I forgot to save and couldn't unearth from Google Scholar. Genuinely surprising considering how useless GPT-3.5 was at the same task.
Interesting. Somewhere I came across https://www.getsmartee.com/voicegpt/# and ever since I am glued. Not sure if this would vanish someday --meaning taken away for some reason. But, I am enjoying it, especially when I am driving and jogging.
Your policy prediction is spot on. The tech majors will pretend no kids use their platforms, and the kids will pretend they're older, and then there will be a panic
Arvind, thank you so much for sharing this. Here is an op-ed my father wrote when I was five about the same topic: https://bit.ly/MacChildPlay92
1. I appreciate your defense of the computers in the upbringing of children, and I think you are right about saying that "screentime" is not a category. I believe a healthy relationship to technology involves some long stretches of "quiet time" — I think Google broke us a bit in two ways: Because we can Google anything, 1) we think all questions are worth asking and 2) we sometimes don't think about questions on our own enough.
2. You are spot on about the risk of addiction when "visual output" is part of the package. I think this is when using the iPhone's grayscale filter will prove useful (bring us all back to the 1990's!): https://www.wired.com/story/grayscale-ios-android-smartphone-addiction/
I think this is great, especially if in the future the parents have more customisability (rather than it being in the hands of OpenAI etc.). In that case, parents might come to trust the AI more than, for example, teachers... as the AI they can programme to be completely inline with their principles rather than those of the ideological teachers (I.E "Come on little Jimmy, don't make a sword from Lego, that's the patriarchy.")
I'd much prefer my kids were with AI whose ideological grounding I programmed.
I wrote about the possibility of these types of use cases a while ago, and now you’ve done it! This moves so fast...
"So far, the bot has not output any harmful content. There have been a couple of borderline cases. One of the stories involved a character dying of old age. Death is not an off-limits topic in our family, so this was not a problem for us."
This does suggest there might be some cases where the bot doesn't say anything *inappropriate* but still might introduce a new idea that parents would want to discuss in-person. I wouldn't want ChatGPT to be the first to tell my toddler that the family dog isn't going to live very long. Religious parents might be especially cautious. But it seems that an LLM whose training data is specially tailored toward children could be an excellent edutainment toy. (Just make sure kindergartners don't ask it for help with math problems...)
I also wanted to say that I really appreciated your recent tweets about using GPT-4 for looking up papers! (I don't have Twitter but read a few accounts.) GPT-4 seems as bad at *reasoning* as GPT-3.5, but it didn't occur to me that it would have a much stronger grasp of its own knowledge base for that kind of search. There are still a lot of hallucinations to wade through, but I finally found a paper about classical conditioning in nematodes I forgot to save and couldn't unearth from Google Scholar. Genuinely surprising considering how useless GPT-3.5 was at the same task.
Thank you very much. Very interesting. I'm sure Generated AI studies in young children's education is an exciting area of research.
Interesting. Somewhere I came across https://www.getsmartee.com/voicegpt/# and ever since I am glued. Not sure if this would vanish someday --meaning taken away for some reason. But, I am enjoying it, especially when I am driving and jogging.
Presumably the age 13 restriction is because of COPPA.
Wonderful work! Bravo!
Great article. Inspiration too. Will try building something similar myself.
Thank you very much for sharing the story and experience - this sounds very inspiring, expanding my horizon with further view points
Wow! So cool. I had a Dad who was like ChatGPT - he knew a little about everything - and conversations with him definitely helped me grow cognitively,
very interesting
Very cool, I love this! 🙏
Really brilliant 👏🏿👏🏿👏🏿👏🏿