Artificial Intelligence-What do YOU think?

drakonius

Chieftain
Joined
Jan 13, 2007
Messages
47
Location
north america
Sometimes my AI acts as if it has some level of real consciousness and intelligence.Once years ago Egypt offered me a page-long treaty offer with all kinds of stuff-a real sweet deal-and it all made sense too. The AI seemed to be rational and aware to some degree. I refused the treaty-then the Americans attacked me and i made a treaty offer to the Egytians(the same deal that they had offered me). The AI as Egypt was offended and it said that i should have accepted its offer when it made it.It then insulted me as if it were a real person\entity and then allied with the Americans in an attemt to destroy me. And often in the Star Wars Galacitic Battlegrounds game units would seem to react with shock and/or fear. I am very big on AI for machines and the last that i read\heard computers are about as intelligent as insects.As anyone who studies insects knows,insects are suprisingly bright-they have small neurons than other animals so they can pack more brain in a smaller space. it is now believed by some scientists that insects are on a par vertebrates in regards to intelligence-even as smart as birds or rodents. Machines are probably "dumb" because they have no experience or programmed instincts. What would a computer or robot with the savy of a real insect be like. This painer once programmed his computer to paint in his style. The computer then went on to paint darn ggod painting that made sense and tha showed artistry. And one scientist once prorammed his computer to ask questions about the world to seek information and to form its own conclusions. It eventually asked him if it was human\a person. What do you make of this? You opinions please.
 
Computers are very good at math, but very bad at dealing with the real world. If you see flashes of intelligent-seeming behavior, give kudos to the humans who programmed and built the program and the hardware. Science magazines have been predicting human-like robots for decades, but they never seem to materialize. Artificial intelligence is still a long ways away, but you'll see good mimicry of intelligence quite often, as programmers get better.
 
You are quite right. There are some darn good programmers out there. the field of AI seems to have been bogged down for the last ten or so years. I heard that they are now looking at neural networks and advanced software.
 
I've had the game do some damn intelligent things, like sneak attack my capital and all cities containing ICBMs - with nukes - then one single frontier city, pouring troops in through the gap and giving me the fight of my life until I was able to get some more ICBMs up and running and take out their industrial base. After that it was just a matter of time, even though they had numerical superiority, and comparable tech. Awesome war, should talk about it in your other thread drakonius.

I've also had Turkistan in TETurkhan's Test of Time scenario leave more than ten cities unguarded, which I proceeded to destroy one by one with my single warrior scouting the Far East. It depends on the situation, and the Civ's traits and preprogrammed conditions I guess.
 
Sometimes my AI acts as if it has some level of real consciousness and intelligence.Once years ago Egypt offered me a page-long treaty offer with all kinds of stuff-a real sweet deal-and it all made sense too. The AI seemed to be rational and aware to some degree. I refused the treaty-then the Americans attacked me and i made a treaty offer to the Egytians(the same deal that they had offered me). The AI as Egypt was offended and it said that i should have accepted its offer when it made it.It then insulted me as if it were a real person\entity and then allied with the Americans in an attemt to destroy me. And often in the Star Wars Galacitic Battlegrounds game units would seem to react with shock and/or fear.

I am very big on AI for machines and the last that i read\heard computers are about as intelligent as insects.As anyone who studies insects knows,insects are suprisingly bright-they have small neurons than other animals so they can pack more brain in a smaller space. it is now believed by some scientists that insects are on a par vertebrates in regards to intelligence-even as smart as birds or rodents.

Machines are probably "dumb" because they have no experience or programmed instincts. What would a computer or robot with the savy of a real insect be like.

This painter once programmed his computer to paint in his style. The computer then went on to paint darn good painting that made sense and that showed artistry. And one scientist once programmed his computer to ask questions about the world to seek information and to form its own conclusions. It eventually asked him if it was human\a person.

What do you make of this? You opinions please.


+=+=+=+=+=+=+=+=+=+=+=+=+=+​

I like what you have to say quite often but paragraph breaks would be muchly appreciated by my old eyes. :scan: This looks like one long sentence as posted but you have interesting things to say.
 
darski,it is true that my posts do run more or less as long single sentences.sorry about that,but thats pretty much how my brain thinks.sorry about the mispells too-I get excited while posting. in my normal writing i am however quite merticulous. thanks for liking my thoughts. what do tou think of machines seeming to show emotion? i would like your words on this matter.
 
I don't know about Civ specifically, but AI in computer games tends to be very much "rules based" - i.e., the computer is just following pre-programmed rules and heuristics. There's no capacity for learning or planning. People are working in these fields, but you won't see it in a computer game for a long while yet.

What you saw was likely programmed in - consider, it's easy for me to write in a program print("You suck"), but that doesn't mean it makes any sense to say the computer was offended, or was really intending to insult you.
 
In fact, there was almost a computer recently that passed the Turing test. So we're definitely getting close to a conversation AI, and from there, we probably wont be that far from true AI in games.

A bit of trivia related to the link Shadylookin posted above. I don't know if that specific article said anything about it, but some judges were actually so convinced by the computers that they were people that they identified some actual people as computers.
 
The turing test is a joke and serious AI researchers know it. Being constrained to a very specific topic leaves the task far easier, and being a chatterbox is far from the only sort of intelligence. Why must an intelligent computer lie about being a human to truely be intelligent?
 
In order to accurately imitate a human in conversation, the computer has to be able to choose what to say according to the questions or statements presented to it. That alone is no small feat.

I for one, am not saying that passing the Turing Test shows that computers are finally intelligent, but it is a stepping stone. It means we've created something that can take what it already has an apply it to different situations without a programmer acting as the puppet master. Take a look at some of the current chatbots that were in the contending for the prize, like Elbot. Spend 30 minutes conversing with it about different topics and you can see thats its not some topic constrained chatterbox. The thing made me laugh several times with its unexpected replies.

So please, don't discredit the Turing Test as some glam. It might not represent true machine intelligence, but at the same time, it's a good stepping stone to something bigger.

Also: Needn't hit "Post reply" 3 times :p
 
The chatterboxes going for these prizes are built to be good at one predefined topic only. While they certainly are put on the web for more topics, that's not how the Turing test is performed today.

Spend any substantial amount of time with any of these prize winning chatterbots with unconstrained conversation and you quickly realize it's a program. It's only when they narrow down to specific conversations can these things do any good.

The big of the problem of the Turing test is there's no good metric for it. How well do these have to perform to win? There's no serious metric given here. Being able to do a single conversation on a single topic under highly idealized rules is basicly what they do know, which is a horsecrap way of measuring anything close to intelligence. On the other end being able to completely mimic a human over years of conversation is probably going to come after developing intelligence.

And in the meantime, developing programs to do better on the Turing test doesn't shed much light unto how to make intelligent machines. This is why noone in AI research takes the Turing Test seriously.
 
I also thought the Turing Test was a pretty decent measurement of AI, better than any other I've ever heard. If a machine can survive 30 minutes (or even 15) of unconstrained conversation with a human who is determined to discover the truth (of whether it's talking to a human or machine) I'd say it's pretty damn impressive. Being able to hold a rapid, complex conversation, conveying information, emotion, being able to switch from topic to topic, respond to subtle cues & learn & use new information is one of the main things that makes us human. If a machine can do this, as well as your average human, I'd say it's pretty amazing (and a bit scary)
 
I did an experiment into the Turing test with a test case, we will call him J.

J: Greetings! Care for some salad? I made it myself!
Me: There shall be peace in our time.
J: Excellent (etc.)
Me: Let's talk about something else.
J: Go on.
Me: What do you think about that guy with the skull on his head?
J: I think it would be better if Montezuma was no more.
Me: Ok, let's talk about something else...
Me: What do you think about Isabella?
J: Isabella? I like him.
Me: Your head would look good on the end of a pole.

I still can't decide if it is a man or a machine ;)
 
Top Bottom