Isn't all implementations of AI a combination of "a simple xyz"?
The perception of Artificial Intelligence being some kind of magic is common. However, as with all magic, when you know how it is done it is "just" a combination of techniques skillfully performed. And with AI it is a combination of algorithms, data structures and computational power with an output that make sense for whatever the application is.
As with other areas of cutting edge research, most non-experts often have little choice but to resort to woefully inadequate simplification. 'Any sufficiently advanced technology is indistinguishable from magic', and all that.
The best way I've come up with to explain the current state of AI is, "we can do tricks by hiding multiple immense databases behind the scenes and having supercomputers run complex searches among them, but we probably won't be able to fully model even a mouse brain for at least 10 years... and at that point we'll likely have learned enough new things about brain function to push it another 10 years out."
To some degree, but if somebody wants to learn about AI in 2016, they probably want to learn current techniques like deep learning. Simple rule based systems like this seems to support aren't going to be very useful to other problems.
I think the grandparent has a valid point. IMHO learning is a necessary feature of any intelligence, artificial or natural. That's why modern AI is 90% Machine Learning. All non-learning approaches (frames, symbolic AI) have failed 20 years ago.
How to make an easily approachable tutorial that combines game-like world and real learning AI (for example Q-learning or maybe evolutionary policy search) remains and open question. Maybe someone will build it.
> 'Isn't all implementations of AI a combination of "a simple xyz"?'
Not really, unless you count mathematical techniques such as gradient descent optimization or convex optimization as simple. In this case a finite state machine sounds very simple in comparison.
Not particularly the type of AI I would teach people (neural networks, and some symbolic AI) but this looks good for working with kids to learn some programming. My grandkids are visiting and I might see if they are patient enough to try this.
do you have any high level thoughts about how those could be used to dynamically solve these puzzles?
I was thinking about how to do this using QLearning to solve the optimal solution given any level, though I am not sure how long it would take to converge since their are so many different actions.
Is there are "tutorial" for this game? This is a really complex topic and if you're trying to teach path finding and other ideas to new programmers it would be really helpful if there was a step by step guide to show the basics of what you're trying to help them accomplish. Something that can get them to the top of the tower, but can also be improved upon. Even if it's just a code sample for each level that won't beat the next level.
a Roguelike dungeon crawler is cool, but i'm not-so-secretly hankering for a procedurally generated text adventure in the style of.. well.. text adventures, requiring programming-orientated input.
I really wish technical people would explain what they mean when they say AI, and quickly name which algorithms they're using. If they don't, it's just clickbait.
The user learns AI by coding. So one can use expert systems, ML, NN, totally random inputs whatever. So the term "AI" in this context seems appropriate.
It's interesting that this style of learning programming fundamentals has taken off such -- Ruby Warrior, The new swift playground announced by apple, and now this Javascript port.
Looks fun but it has little to do with AI, I expected rather something like https://top-tactic.com where you can train NN how to play football just by creating sample situations on the pitch.
32 comments
[ 3.6 ms ] story [ 71.7 ms ] threadThe perception of Artificial Intelligence being some kind of magic is common. However, as with all magic, when you know how it is done it is "just" a combination of techniques skillfully performed. And with AI it is a combination of algorithms, data structures and computational power with an output that make sense for whatever the application is.
The best way I've come up with to explain the current state of AI is, "we can do tricks by hiding multiple immense databases behind the scenes and having supercomputers run complex searches among them, but we probably won't be able to fully model even a mouse brain for at least 10 years... and at that point we'll likely have learned enough new things about brain function to push it another 10 years out."
How to make an easily approachable tutorial that combines game-like world and real learning AI (for example Q-learning or maybe evolutionary policy search) remains and open question. Maybe someone will build it.
Not really, unless you count mathematical techniques such as gradient descent optimization or convex optimization as simple. In this case a finite state machine sounds very simple in comparison.
I was thinking about how to do this using QLearning to solve the optimal solution given any level, though I am not sure how long it would take to converge since their are so many different actions.
Would the longer second line in the instructions
really have scared off that many people?Both are AI games where you code your own player class, but MIT competition is PVP(player-versus-player) whereas in this one you play alone.
Although I would argue he found a better use for it.
There are very few cases when games have used interesting AI/Machine learning techniques but most of the time you are safe assuming that they are not.
I've gotta take the opportunity to plug the interactive version of Ruby Warrior here: https://www.bloc.io/ruby-warrior
It's interesting that this style of learning programming fundamentals has taken off such -- Ruby Warrior, The new swift playground announced by apple, and now this Javascript port.