Anyone taking an AI class?

I'm currently taking AI as a undergrad level class. The book I'm using is Intelligence: Structures and Strategies for Complex Problem Solving by George Luger 5ed.

 

Log in or Sign up to hide all adverts.

I'm in the 5000 level (Master's level) Advanced Artificial Intelligence course right now. Not really my field of study, but can definitely be some interesting stuff. I'm not sure what book they're using, we don't reference it much. My professor usually just references research papers for anything truly novel.

 

Log in or Sign up to hide all adverts.

Log in or Sign up to hide all adverts.

Yeah, because we know that anything that doesn't yield instant gratification or otherwise tangible results is a complete and utter waste of time.

There's no such thing as studying something for interest in personal development, or otherwise a better understanding of anything.

 

I agree with gukarma here. It is a valid field of study even if it had NOTHING to show for itself. The field of computers in general had almost no immediate benefits, they were too big and almost useless... but that changed.

As for AI in general, myself and most people SERIOUSLY looking at the field believe that we are many years from any sort of true AI. The biggest problem with AI is in the description of the problem; how do you define true AI. For years people said... "If we can teach a computer to play Chess better than a human, that's AI". Now that we have, many simply call it a neat trick, and realize its mostly heuristics and brute force. Others have said that if we can get a computer to interact with a human in a useful manner, that's AI. Well, call and order plane tickets from Delta (I believe that's the company) and there is a good chance you'll be talking to a computer, not just punching keys, but actually talk to the computer. People no longer consider it AI. Now people are saying if we can crack the machine translation problem, THAT will be AI. People are saying if we can get a car to drive autonomously, THAT will be AI. But of course, as these things are coming to fruition, they are demoted to simple engineering feats, rather than the emergence of AI.

We still have the Turing test, but I believe that someday we will achieve a computer that will pass the Turing test and people will still not consider it AI because it will still use some tricks. They'll call it clever engineering and discount it. The biggest problem is that people consider AI to be a form of mysticism and as soon as they know how it works, it can't be mystic anymore, and therefore can't be true intelligence.

-AntonK

 

I think the answer to this question is because it has almost NOTHING to do with AI, at least the concept of true, pass the Turing test, converse with you AI. The types of problems you are talking about don't necessitate AI at all, they necessitate rather complex systems and some clever engineering, but not AI. Many of those problems are already being attempted with other machine learning techniques. The point for me, is that if you are building a system to tackle a specific problem like that, then it isn't artificial intelligence, its simply another extremely useful computer program.

 

I study neural networks at uni. I saw a cool video at the end of one lecture of these 'insects' some guy built. Their brains are made completely from artificial neurons are they move about like insects on six legs. Hold one leg still... and it'll learn to use the other legs independently.

But I don't think this is anywhere near the level of AI that scifi fans want.

operating software? what do you mean by that - operating system? or just a program that has some desires, can learn and use new data and techniques to accomplish its goals?

The hardware requirements for a neural based computer:- billions of artifical neurons (a human brain has 100B, but our brains have other functions than just consciousness) that can operate at a reasonable speed (0.001 seconds switching time for biological neuron). Not that speed is strictly an issue of intelligence.

For a conventional computer: it would depend on the implementation type you are working with and how 'clever' you wanted it to be.

Sources: me and a pdf found on google with some basic facts

 

Your question is like saying, "Are you taking human biology? If so, when can you guys clone humans? If you don't have an answer to that, you're wasting your time."

 

'True AI operating software' is not a phrase I would expect to hear. Relentless arugments about 'what is inteligence' is what you will find if you talk to someone who knows what they are doing.

 

and that's what I have to say about your initial comment

Please Register or Log in to view the hidden image!