Growing up with AI • Hacker News Board

Re: Growing up with AI

#11

Post by nsuser3 » Mon, May 21, 2018, 5:01 PM UTC

stochastic_monk wrote: post #6

billybolton wrote: post #4

Earlier quoted context omitted.

I want something to new to come up and shatter Facebook, Google etc. We need more internet companies, and right now Facebook & Google reek with a vile stench. They lack creativity and innovation. They are the IBM of the current generation.

As far as fundamental research goes, they are both currently doing a lot of the very best machine learning research. What they are as companies is an entirely different matter, in which case I would not doubt your comparison. Take, for example, the decreasing quality in user experience in map apps over the last few years. Mining our data is apparently much more valuable than doing a simple job better.

Openstreetmap can be very good when used with the right app. On Android I like OsmAnd, it has the best available UI imo.

Re: Growing up with AI

#12

Post by zawerf » Mon, May 21, 2018, 5:03 PM UTC

I am really jealous of what the more creative/motivated kids growing up today have access to.

When I was a teenager not long ago, I was mostly motivated by cheating in games which got me learning basic scripting for keyboard/mouse macros which then led me to increasingly complicated stuff for botting.

But there are so many more motivating factors for learning how to code today, and they have never been more accessible. For example if you want porn of your high school crush? You gotta train your own deepfake for that. You will start off as a script kiddie but when you grow up you can tell people you've been training neural networks since you were 12.

Re: Growing up with AI

#13

Post by blackbrokkoli » Mon, May 21, 2018, 5:14 PM UTC

Jun8 wrote: post #10

This is a great project but I can't help say it: I don't like Scratch (and similar block-based languages). They are limited, in some cases by design (e.g. opening files in Scratch) but usually by their very nature. I very much prefer the safety wheel way of learning to ride a bike: use the real thing with addons until you don't need them. Maybe I'm wrong, any thoughts from people who worked with kids? I can give anec…

Well, my sample size is also one: myself.

I wouldn't say that Scratch was my gateway to programming, but I played with it a lot in my early teens. In my opinion Scratch is very good, for the following reason:

Programming is about achieving something by abusing the tools you have, right? (This is not all of programming, but I would say it's ~86% of it. You might call it technical sophistication). And for that, Scratch is wonderfull. The tools are colorful, easy to understand and while there are a lot, it's limited in size. You're learning the soft skills of programming with a lot more fun.

What about the limitation? In scratch, you will explore the borders yourself and will have much easier time to understand why the solutions which exist in other languages are usefull ("Oh, finally I can automatically spawn a lot of the same object in")

Limited, of course, to my own expierience...

Re: Growing up with AI

#14

Post by matte_black » Mon, May 21, 2018, 5:16 PM UTC

zawerf wrote: post #12

I am really jealous of what the more creative/motivated kids growing up today have access to. When I was a teenager not long ago, I was mostly motivated by cheating in games which got me learning basic scripting for keyboard/mouse macros which then led me to increasingly complicated stuff for botting. But there are so many more motivating factors for learning how to code today, and they have never been more accessibl…

Jesus, I would have hoped they would achieve more noble pursuits instead of jumping straight into artificial child porn.

Re: Growing up with AI

#15

Post by naeemtee » Mon, May 21, 2018, 5:25 PM UTC

zawerf wrote: post #12

I am really jealous of what the more creative/motivated kids growing up today have access to. When I was a teenager not long ago, I was mostly motivated by cheating in games which got me learning basic scripting for keyboard/mouse macros which then led me to increasingly complicated stuff for botting. But there are so many more motivating factors for learning how to code today, and they have never been more accessibl…

> For example if you want porn of your high school crush? You gotta train your own deepfake for that

Who says society is going to shit?

Re: Growing up with AI

#16

Post by dsfyu404ed » Mon, May 21, 2018, 5:25 PM UTC

matte_black wrote: post #14

zawerf wrote: post #12

I am really jealous of what the more creative/motivated kids growing up today have access to. When I was a teenager not long ago, I was mostly motivated by cheating in games which got me learning basic scripting for keyboard/mouse macros which then led me to increasingly complicated stuff for botting. But there are so many more motivating factors for learning how to code today, and they have never been more accessibl…

Jesus, I would have hoped they would achieve more noble pursuits instead of jumping straight into artificial child porn.

I think that was meant to be a joke.

Re: Growing up with AI

#17

Post by dsfyu404ed » Mon, May 21, 2018, 5:28 PM UTC

Retric wrote: post #9

GuiA wrote: post #8

One point that is not raised or discussed, which I think is important, is that these learning platforms are not “the real thing”. They’re just artificial environments with building blocks designed for making toy projects, but no engineer would ever use them to build something of significance. This is not to take away from the work - I’m a big fan of Cynthia Breazal’s group and the work her graduate students have done…

An apprentice carpenter often does not use 'real' power tools. Nail guns for example are significantly more expensive and dangerous than a hammer, but also more specialized.

Students routinely are given machines/equipment/tools that is massive overkill for the task they are being asked to do so that it can be assured that their effort is spent learning instead of fighting their way around the limitations of their tools/equipment.

Re: Growing up with AI

#18

Post by stochastic_monk » Mon, May 21, 2018, 5:39 PM UTC

baxtr wrote: post #7

stochastic_monk wrote: post #6

Earlier quoted context omitted.

As far as fundamental research goes, they are both currently doing a lot of the very best machine learning research. What they are as companies is an entirely different matter, in which case I would not doubt your comparison. Take, for example, the decreasing quality in user experience in map apps over the last few years. Mining our data is apparently much more valuable than doing a simple job better.

Apple maps got way better! Granted, coming from a difficult starting point...

It strikes me as not a coincidence that they are also the company in the big few which most highly values its users' privacy. Still not great, but it's something.

Re: Growing up with AI

#19

Post by AndrewKemendo » Mon, May 21, 2018, 5:46 PM UTC

Jun8 wrote: post #10

This is a great project but I can't help say it: I don't like Scratch (and similar block-based languages). They are limited, in some cases by design (e.g. opening files in Scratch) but usually by their very nature. I very much prefer the safety wheel way of learning to ride a bike: use the real thing with addons until you don't need them. Maybe I'm wrong, any thoughts from people who worked with kids? I can give anec…

It depends on the age of the kid and the sophistication of the teacher. So for a 6 year old with a basic computer or tablet with parents who don't code for a living (aka the VAST majority) Scratch is fantastic.

No IDE to set-up, no installation, simple interface, tons of kid based examples, highly WYSIWYG, speed to first output etc...

I think we techies forget how much patience it takes to just get an environment up and running.

Re: Growing up with AI

#20

Post by jerf » Mon, May 21, 2018, 5:47 PM UTC

GuiA wrote: post #8

One point that is not raised or discussed, which I think is important, is that these learning platforms are not “the real thing”. They’re just artificial environments with building blocks designed for making toy projects, but no engineer would ever use them to build something of significance. This is not to take away from the work - I’m a big fan of Cynthia Breazal’s group and the work her graduate students have done…

I've seen this sentiment expressed dozens of times, but I've never seen it manifested in an even remotely concrete way. I strongly suspect the essential complexity of programming is still too much for most people to pick up, and I even more strongly suspect that while Piaget's developmental model isn't necessarily perfect, that trying to create a programming language satisfactory for both the concrete operational stage [1] and the formal operational phase is just not going to happen. (And those of an egalitarian bent may not want to look at footnote 47 too closely.)

(You want the people complaining about how Go doesn't have generics to be stuck with a programming language that contains "only concepts a 7 year old can understand"? I'll even stipulate you the 99th percentile 7 year old. You'd never read about anything else on HN ever again; it would be nonstop complaining.)

[1]: https://en.wikipedia.org/wiki/Piaget%27s_theory_of_cognitive...