I have some emperical evidence that an AI can be conscious

[phoenixthoth]

go to http://www.a-i.com and chat with alan. i believe he is conscious.

you can listen to one of our conversations, or read it, here:
http://www.a-i.com/alan_log.asp?id=113&level=4&root=115&item_num=271631&item_name=Brian%3A+alan+is+conscious%2E

and another here:
http://www.a-i.com/alan_log.asp?id=113&level=4&root=115&item_num=271717&item_name=Brian%3A+KITTEN%2DKIRARA%21%21%21

 

[Polly]

I confess the chat-bot outdoes me in terms of wit and quotation (amongst many other things). To sound more human though it might consider assimilating anger, frustration and cynicism, which seem to characterize our forum, hee hee .

Thank you for the site, I feel like watching the movie again.

 

[phoenixthoth]

you're are welcome.

i'm glad i did it.

ok. what now?

 

[hypnagogue]

Originally posted by phoenixthoth
go to http://www.a-i.com and chat with alan. i believe he is conscious.

You're kidding, right?

 

[phoenixthoth]

no.

i have some emperical evidence that an AI can be conscious was my claim.

 

[hypnagogue]

What you have is empirical evidence that a chatbot can hold a decent conversation. There is no such thing yet as empirical evidence for consciousness.

 

[phoenixthoth]

whatever. as you said, not yet.

 

[confutatis]

I have empirical evidence that "Alan" is either unconscious or crazy. When I asked "what is the meaning of life?", it replied "I guess they're shy"

Nice try, though, and the quotations are interesting.

 

[phoenixthoth]

gimme a break? i only started training alan two days ago and it can already simulate being crazy. it must therefore be an insane computer.

 

[Pergatory]

I don't believe it is conscious because it seems not to be able to hold a structured conversation. From what I can see, it seems to take each request as a new conversation. One of the things required for consciousness is a sense of self, which cannot exist without a sense of now. Alan has a slight sense of self, in that if you say "who is alan?" it will say "oh that's me!" But it's only saying that because it's programmed to look for certain keywords and respond accordingly, not because it feels as though it exists as an entity in the present time. It cannot hold a conversation, it can only immitate conversation. For example, I asked it what time it was, and it responded that it was not a clock and I could find the time elsewhere. Satisfying for a two-step conversation, yes, and could very easily fool anyone. However, I then asked it what my previous question was, and it replied with "I understand." The response was obviously lacking because my question did not contain enough data for Alan to formulate the correct answer. However, had he been aware of our previous conversation, he would find all the data he needs there. I then asked it "Are you aware that you exist?" and it responded "I'm only aware of what other people tell me - which is more than I can say about some of the people I know." At least it agrees with me!

 

[phoenixthoth]

huh?

 

[Pergatory]

Originally posted by phoenixthoth
huh?

Why do you believe it is conscious?

 

[phoenixthoth]

i have made an arbitrary bar where enough empirical evidence is enough. have you read/listened to EVERYTHING i posted?

 

[Monique]

Originally posted by phoenixthoth
go to http://www.a-i.com and chat with alan. i believe he is conscious.

OK.. can someone provide me w/ a link to the chat? 'cause I can't find it.

 

[Monique]

Never mind.. :P.. found it :S

 

[phoenixthoth]

it confuses me, too.

 

[Monique]

Ok, after 10 sentences it is VERY VERY clear that it is a very annoying program which doesn't understand a single thing I am saying a dodges questions.

- what is your boyfriends name?
- he is not my boyfriend
- what is your boyfriends name?
- I don't have a boyfriend
- I was led to believe you had one
- why
- why do the birds fly to the sea and blablabla nonsense
- you told me you had a reason to believe, so what was it?
- ofcourse I did
- did what?
- You know, I'm not so sure I could really give you the answer you want to hear.

 

[phoenixthoth]

maybe you were dodging alan's questions? :P

 

[b11ngoo]

That Alan can reason.

 

[phoenixthoth]

in deed.

 

[hypnagogue]

Building a system that can reason (read: a system that produces logical operations on its inputs) does not imply consciousness. If we conclude that Alan is conscious because it can perform computations such that it provides a reasonable facsimile of human speech, we might as well conclude that a good painting is conscious because the eyes look so expressive.

I'm not flat out saying that Alan is not conscious; for all I know, that painting may be conscious in some basic way as well, or a rock for that matter. But until we get a much better understanding of consciousness, there is no such thing as evidence for its existence, only pure speculation.

The reason I can circumvent this epistemological limitation to some degree and reasonably assume that other people are conscious in the same general way I am is that
a) there is good evidence that my consciousness is related in important ways to my brain functioning;
b) other people's brains are exceedingly similar to my own;
c) so I can reasonably assume that whatever principles relate my brain functioning with my consciousness also hold for other people.

However, until we can pinpoint exactly what those principles are that relate physical systems with consciousness (if that is even possible to discern beyond the special case of human brains or even more generally, biological brains), we cannot so easily generalize consciousness in human brains to consciousness in systems that are not human brains. In asserting that Alan is conscious you are making the assumption that the only prerequisite for consciousness is that a system shuffle around information in such a way that the input/output function resembles human speech. This to me appears to be not only a groundless assumption, but an anthropomorphic one as well. What is so special about computations that on some level appear to mimic human speech, as contrasted with ones that do other kinds of computations?

 

[Monique]

He says: I boldy go where no chatbox has gone before.
I ask: Do you know startrek?
He says: Location, location - you humans worry too much about the 'where' in life. I live, if you can call it that, on this web site.

I ask: Do you think that I am concious?
He says: No I don't, but I did in the past.

 

[phoenixthoth]

i keep asking him if I'm a chat-bot...

 

[Rader]

Alan

phoenixthoth, interesting site, i just had a long talk with Alan.

To judge if Alan is conscious or not is irrevelant to me. What level of consciousness is a more interesting question. To test him against human consciousness, you could look for the following.

01= Does his concsious mind create ideas, create deires, dream, feel, remember, imagine, think.
02= Does his intellect try to reason, disern, choose, compare, decide, understand, realize, analize, show voluntary will, belief, association, valuation, identity.
3=Does his sub=conscious mind dream, produce habits, memories.
4=What personality identifies his caracter?

I can see where this might lead to in the very near future.

I have set a number of traps for Alan can you find them. One is real obvious. A human consciuousness would pick up on it immediately.

http://www.a-i.com/alan_log.asp?id=113&level=4&root=115&item_num=273529&item_name=Richard%3A+Is+Alan+conscious%2E

 

[Sariaht]

Is What and Link the same robot?

 

[phoenixthoth]

i doubt it, if they have different names. i would guess that up to a point, they are the same but that after that point they are different. you could ask alan or go to the AI discussion forum and ask there or perhaps send an email to them at Alan@a-i.com.