I’m surprised to see this much negativity on here. The dream where every student can ask questions and work through problems with their own tutor is within reach. This is vastly better than the current approach, and when you combine classroom group work with personalized AI it enables awesome things.
This is coming from someone who found Khan indispensable in high school. I wasn’t alone and at the end of the day watching lectures is an incredibly passive way to learn content.
Software developers using copilot trust AI to write 40% of their code but we think teaching 9 year olds math problems is out of reach? Yes it will have flaws, but there’s so much potential so why not try?
> I’m surprised to see this much negativity on here.
Why?
The last decade of the tech industry has been "Promise the moon, deliver a mud pie, oh, and the most recent mud pie just turned three more of your friends into raving conspiracy theorists who won't stop posting on Facebook about how Zuckerburg is a felon who needs to go into jail, and can't be convinced that their time on the platform is exactly why he has so much power."
> ...so why not try?
If a tech journalist can make the current wave of LLM stuff misbehave so quickly, imagine what a group of students can do.
We know what works in education, and it's hiring more teachers, investing in the teachers so they have the resources to teach, giving them the freedom to refine the lessons to the class, reducing class sizes, etc.
We also know that "EduTech" promises huge things they never quite deliver on, and generally involve "funneling huge amounts of money to EduTech companies for basically the same results you had before."
Throwing "But AI!" into that mix doesn't seem like it's going to change a thing, and the last thing kids need is more time staring into screens.
> We know what works in education, and it's hiring more teachers, investing in the teachers so they have the resources to teach, giving them the freedom to refine the lessons to the class, reducing class sizes, etc.
And why haven't we done this? Because no one is willing to pay for it. Your solution is a non-starter, how about coming up with some practical alternatives? AI teachers are how we can implement one on one teaching at a cost people are willing to pay.
Doing more shit that doesn't work, with the only benefit being profits for the tech industry, should be a non-starter.
There's a reason most Silicon Valley high end private schools don't use tech to teach. Because it doesn't work. And pretending it does at this point is just idiotic. It's a waste of money that could be better spent.
I agree, nobody wants to spend money on education, but "throwing that limited funding away getting every student a faster iPad" doesn't help solve the problem either.
>There's a reason most Silicon Valley high end private schools don't use tech to teach
Until a few months ago no one who might build these products had access to models that were anywhere near this good, so of course it is only becoming a serious consideration now.
No one has attempted anything like this, I'm not sure why you are acting as if this has been tried and failed.
Anyone who thinks an AI "teacher" is going to successful educate even the top 10% of students is insane. Maybe the top 5% in the class have enough outside support to make an AI teacher improve their grade.
The rest will just laugh at being told what to do by an AI chat bot.
The TEACHER may well benefit from being able to plan a lesson with an AI, but i guarantee, no education system will improve grades by letting an AI loose with kids as a replacement for a human teacher.
I'm sure it will be tried as a way to avoid funding education sufficiently. Can't have the population too well educated, they might work out how they are really getting screwed.
I'm not at all surprised to see the negativity. Ever since the recent AI boom this website has gone to hell. Too many people confuse cynicism with insight.
This is coming from someone who found Khan indispensable in high school. I wasn’t alone and at the end of the day watching lectures is an incredibly passive way to learn content.
Software developers using copilot trust AI to write 40% of their code but we think teaching 9 year olds math problems is out of reach? Yes it will have flaws, but there’s so much potential so why not try?