How to train a computer to think like a human?

Home » Programming & Design » How to train a computer to think like a human?
Programming & Design No Comments

Best Answer:

RAGHAVENDRAN: Can *I* train *YOU* to think like a programmer?

This programmer's way of thinking is to ask you to elaborate. Your question is a bit ambiguous.

Are you asking how *YOU* can train a computer to think like a human? Or are you asking how computer scientists with doctorate degrees are trying to train a computer to think like a human?

Or, are you just asking about artificial intelligence, which tries to find more efficient methods where a computer can find patterns in data to answer questions. For instance, the IBM Watson computer is being used to make medical diagnoses. And, because it is NOT thinking like humans, is providing answers that doctors have missed.

Human think the way we do because of evolution. And, because we have senses. We have opinions on what food tastes better than others. However, we don't all agree on that. I hate cheese. I don't hate the smell of coffee but I never drink it. I prefer cold beverages. Even though coffee is free at work I have NEVER had a sip of coffee. I could drink iced coffee, but, it wasn't generally available to me, so I never got used to the idea. I think I have accidentally had a Jamoca shake at Arby's, and I finished it, but I wouldn't ever choose that flavor when chocolate is available.

I have absolutely no interest in sports. Why? I don't know? I don't understand why anyone is interested in sports.

So, what does "thinking like a human" even mean?

Other answer:

RAGHAVENDRAN:
How do you train a human to think like a computer?
dday9:
In simple terms you create a neural network to learn how to interact and then use a genetic algorithm to breed a population which gets better after each interaction.

This is difficult, if not impossible, to mimic the behaviors a human. If you want to teach a computer to do a specific task or interaction then it is entirely possible to do so, in fact you see this a lot in chess CPU's.

I've included a link in the sources showing how one programmer applied this concept to Super Mario Bros.

Ian:
You cannot train it, you have to make an AI that can LOGICALLY learn from the way humans act from its own deductions and observations, otherwise it'll be more fallible than a normal person. Because unlike computers, people can see inconsistencies and accept them if they were indoctrinated. however if a computer is given a problem (situation) where what it was "taught" isn't actually correct or applicable it will freak out or crash unlike us who will continue with acting out what we were taught against logic.
husoski:
Ever read the news? Schools and education are topics that start arguments and even wars world-wide. So, it seems we don't even have a firm grip on how to teach a human to think like a human.

Simulation is, I believe, the closest we'll ever get. If it takes more than one neuron to "understand" the function of a particular neuron in the context of a brain under study, it surely must be impossible for any brain to understand itself. Well, any *finite* sized brain anyway.

Chris:
You can't.

Only a brain can think like a human, by definition, so we'd need a brain simulation, not computer code.
They are working on that though (the artificial brain).

Chau:
I think you need to write a program that can do that work. You program must be intelligent enough to recognize and train the computer.
Hego:
Artificial Intelligence.

LEAVE A COMMENT