Hacker News Reader

Comment by robocat | Hacker News Reader

How much has our knowledge of AI training techniques helped to discover how to train people to think better?

We've had knowledge of how to eat better to not get extreme scenarios like obesity and look at the effectiveness of that. Until you have a pill that makes you think better only the motivated will do it, and in this case the motivated could already do it.

You seem to imply the 'motivated' can not improve.

I'd say the motivated often reap the rewards of innovations more so than the average, as they were pushing the boundaries in the first place.

Having a dishwasher or a robot vacuum does not make me lazy. It allows me to do more productive things.

I guess I was focusing on the most obvious cases. If a pill makes everyone skinny you'll notice much more that there's no more huge people than that skinny people's body fat went from an average of 13% to 11% or some such. I do agree the motivated have more likelihood to improve, but on a societal level I tend to think first of raising the bottom of the group. I agree with your points.

As someone with an educational background I actually often ask myself the opposite: Why don't AI techniques almost never seem to use the knowledge we have about human learning to train better AI?

Maybe an area worth exploring, if you think there's something to it!

Wasnt that exactly what they tried to do for the first 30 or so years, but it wasn't very fruitful?

So far, I don't think we found anything interesting, yet.

Reply on news.ycombinator.com