For a laugh, yes sure do that lol.

Theoretically LLMs could be great as digital life assistant. IF there wasn't a malicious corporation behind them. Like if you had an open source, locally run AI model that slowly gets to know you or your kids, remembers all the conversations, in it's own encrypted storage that only you (or your kid) can access using a password of voice footprint. As AI models become smarter they could increasingly help manage your life. Or offer tips and coaching about your issues. Even positive reinforcement from an AI could help people grow as people - theoretically.

Obviously you don't want a sycophantic LLM induce narcissism in your kids lol. But if I consider my confused childhood I can imagine some additional attention by an AI with broad knowledge could definitely have helped me.

Currently the attention of kids is largely influenced by algorithms and what is trending. The semi-intelligence of LLMs could provide a counterweight to that. There are studies that just altering the order of search results can change the outcome of political elections. The same is true for a million other things, just for maximizing ad revenue. Again local LLMs could help research and counteract such malicious influences. Instead of recommending a search result an AI agent could look up who owns the site, what history they have, how many falsehoods they published and what political messaging or harmful or deceptive propaganda they try to push when analyzing their talking points.

PS: Obviously you couldn't trust an AI corporation like OpenAI with this, but you could use a modified open weights model for this and a non-profit financed by individual donations.