Considering the massive amount of resource expenditure that it has taken to develop LLMs, is this really cheaper than just giving kids a tutor?
I work in adult education, at a company that services government contracts. We keep hearing about how our clients are dreaming about AI making the "perfect" learning experience for their employees, customized to their every need.
Or at least, that's what they say. They say they want customized experiences for students, but what they really mean is they want to cut costs. Because if it was just about making good training, they could just keep paying us, stop trying to nickel-and-dime us on everything, and have customized, personalized training experiences because that's what human teachers already do.
So they don't want to pay anymore and are hoping AI is going to do it for them. They ask us about it and we talk about what it would take to even make a pilot program that may or may not get anywhere close to their goals (honestly, probably not. I've done some tests and it makes for a good sales demo, but actually making a good quality, consistent product out of it is a whole research project on its own). And the cost is... tech project costs. And they don't want to pay.
In other words, human teachers aren't actually that expensive.
I work in adult education, at a company that services government contracts. We keep hearing about how our clients are dreaming about AI making the "perfect" learning experience for their employees, customized to their every need.
Or at least, that's what they say. They say they want customized experiences for students, but what they really mean is they want to cut costs. Because if it was just about making good training, they could just keep paying us, stop trying to nickel-and-dime us on everything, and have customized, personalized training experiences because that's what human teachers already do.
So they don't want to pay anymore and are hoping AI is going to do it for them. They ask us about it and we talk about what it would take to even make a pilot program that may or may not get anywhere close to their goals (honestly, probably not. I've done some tests and it makes for a good sales demo, but actually making a good quality, consistent product out of it is a whole research project on its own). And the cost is... tech project costs. And they don't want to pay.
In other words, human teachers aren't actually that expensive.