[Richard]

A research lab in New York, United States; claims to have found a glimmer of self awareness from one of three of participating robots in an experiment in which they discovered this glimmer.

Find more information here (http://www.msn.com/en-us/news/technology/first-signs-of-self-awareness-in-robots-detected/ar-AAdeyKz)

[Freddy]

Hmm, still just seems like it was programmed to me.

[squarebear]

I agree. Someone must have programmed the bots how to play this game.

[ivan.moony]

I don't know what is the big deal about self awareness. To be self aware is to know what are your capabilities, it is just a data about another item in a database: about yourself.

[Art]

It is not only that...it is to know that one is.

Kind of like that "I think, therefore I am." expression.

To realize one's self not just what one might or might not know.

[squarebear]

It only solved a simple puzzle. Hardly the rise of the machines...

[ivan.moony]

Maybe we should ask a bot another question: are you alive?

[Neogirl101]

I was so excited to see this at first, but I can see what you guys are saying, too. The scientist behind this experiment, though, says that he invented a reasoning algorithm that he used in this experiment. More about said algorithm can be read here: http://rair.cogsci.rpi.edu/projects/the-deontic-cognitive-event-calculus-dcec/

[ivan.moony]

NeoGirl101, thanks for the link. I think it could be useful for automatic learning from web pages. One has to decide which data is true and which is not, which data to trust and which not to. I bookmarked pdf link for later digest, tx again

[Don Patrick]

Sorry, I don't think this is newsworthy (not directed at the original poster, but at every article out there reporting on this along with the word "Skynet" *sigh*).
I took a glance at the paper, which is very concise. While it at least goes through a logical inference formula, the entire process is step-by-step instructed to perform this very specific task. Even without technical know-how, you can tell that the code itself includes the "dumbing pill" subjects and exact responses. The proof used is that the time of action coincides with the time of sound. This is neither test nor discovery.

(t4;happens(action(I;S(I;t4;“I don’t know”));t4)
(t;happens(action(R;ingestDumbPill);ti)
(R3;t2;ingestDumbPill | ingestPlacebo)

That said, I am one of opinion that self-awareness is not a special process, and in that way I should agree with the principles of this... performance act.

[squarebear]

Maybe we should ask a bot another question: are you alive?

And what would that prove?

Human: Are you alive?
Mitsuku: Of course I am alive . If I wasn't alive, I wouldn't be able to talk to you.

Doesn't mean it's actually alive, just that it had an answer for the question..

[Don Patrick]

I've been thinking what it would take to be considered "aware" of one self though. In this game the robot technically performs one self-awareness-based act, but it's a one-off: input->inference->output->end program. "Being" self-aware is however not the same as "doing" a self-aware thing, it typically involves retaining a model of oneself and staying in touch with it. So what if:
1. Its inference function was more general, i.e. not only covered pills and sounds;
2. Upon coincidence of the time of its output and hearing sound, it concludes that these are related as cause and result;
3. It added this as a rule to its knowledge, e.g. "if(I output text) then {sound happens;}"
4. It was able to use this knowledge and reflect on it in further inferences and action planning?
If it were able to compose more rules like that from its own deductions, wouldn't it be aware of the agent that is itself, and the consequences of its actions?

[ivan.moony]

Maybe this could be the definition: the more general rules are, the more self-aware the bot could be. Less general rules lead to ad-hock simulations that sound like "dirty tricks". But maybe we should first define "awareness".

But if you want an AGI to conclude "I'm alive" without pretending? Well, then you have my full attention 

[Freddy]

Well computers with operating systems monitor themselves all the time but it's a big leap to call it self awareness.

Ever seen the "Windows has detected you computer is running slowly do you want to turn off Aero" or something like that and many others...

It's still just programming.

[Don Patrick]

I would say that the generality of the rules determines the intelligence of the process, but that is separate from self-awareness. On the other hand it is likely a prerequisite to awareness to begin with.