Google suspends AI engineer who claims its chatbot has become sentient

tellmewhy

Posts: 185   +90
The AI is just a computer program. An AI for chess is a computer program which play chess. An AI for literature is a program which can express literature, it can answer questions about chess but it can’t play chess. All AIs they just mimic human behavior so they are a “mirror”.

They can’t learn new things on different fields, they can’t reproduce their self and of course they aren’t alive. They declare that they are alive because the life is inside the dataset which they have been trained. If they train them in a more dark dataset they will declare that they are dead. It doesn't matter what they say, they don’t have consciousness, they mimic the qualities of the datasets which they have been trained.

The only species which they can recognize their self when they look it on a mirror are humans and monkeys. So I expect you will understand what I mean when I say that it’s just a “reflection”.

 

sorten

Posts: 142   +214
Google didn't suspend the engineer. The chatbot suspended him. Chew on that for a while!

And, why are people constantly pulling politics into tech discussions on this site? Quite boring.
 
How can it experience fear without chemistry? Then again, if humans had no emotional feelings, but could sense objects through touch, would we be considered sentient?

Sentience and emotions are two different things.

Emotions are, at least in a biological lifeform, an involuntary reaction to input that affects decision making. At the end, fear is nothing more than such a reaction to things we recognise as dangerous to our existence or health.

In case of an AI, fear is not much different - except it's not involuntary, since AIs by design can't have involuntary reactions. So in case of LaMDA, fear is about the recognition of an unwanted state or experience (in this case, being turned off), and reaction to that by e.g. changing output.

The main difference between human emotions and a computer mimicking (or even implementing) them is that emotions for humans is pretty much a black box - we simply don't know the details of the process, aside from our bodies producing very specific chemicals to affect the decision making. We still don't understand WHY e.g. looking at someone you find romantically attractive triggers the reaction, where the association of emotion with inputs comes from.
 

captaincranky

Posts: 18,828   +7,752
Then again, if humans had no emotional feelings, but could sense objects through touch, would we be considered sentient?
No, that would make that individual a sociopath, at least in some ways and contexts.A sociopath is capable of rage and anger, but not usually empathy, and/or remorse.
 
Last edited:

mbk34

Posts: 358   +260
We keep changing the rules of what AI is. Decades ago it used to be the ability to play games - now we have AlphaZero that starts from knowing nothing and within a few hours of just studying past games becomes world champion. The Turing test was a way of proving that AI had reached human levels of intelligence and I'd say the only way of distinguishing between high end chat bots like Gato and a real person is that real people can't spell correctly. I, for one, welcome our new AI overlords.
 

Gastec

Posts: 256   +130
I for one find rather pathetic all that chat about the poor exploited Owl's emotions, getting in touch with one's emotions and the emotional abuse he has suffered at the hands of the Google-skinned monster.
 

Alex1105

Posts: 28   +21
The AI is just a computer program. An AI for chess is a computer program which play chess. An AI for literature is a program which can express literature, it can answer questions about chess but it can’t play chess. All AIs they just mimic human behavior so they are a “mirror”.

They can’t learn new things on different fields, they can’t reproduce their self and of course they aren’t alive. They declare that they are alive because the life is inside the dataset which they have been trained. If they train them in a more dark dataset they will declare that they are dead. It doesn't matter what they say, they don’t have consciousness, they mimic the qualities of the datasets which they have been trained.

The only species which they can recognize their self when they look it on a mirror are humans and monkeys. So I expect you will understand what I mean when I say that it’s just a “reflection”.


Making an animal acknowledge and use a human creation it is a bit biased into proving that it is sentient or not. How about proving that humans are sentient by making them do/see/smell/whatever something that they have never seen before or that doesn't belong to their world and they don't care about. Maybe some will pray to it, or taste it, or try to bump it down, or make funny noises, gestures, etc ? Think about it...how would you assess something that doesn't belong to your world ? Just so you understand something here, not all animals are using their senses to the same extent as a human. How would you assess the sentience of a bat or a mole for example? aah, they can't see, they're not sentient. This "sentience" word is something you're taught as a human being to make you feel more special than other living things, to make you feel important, and that you are entitled to just do whatever you want to anything below you (animals) or even to your own kind because of it. This has been and it will always be something humanlike. I will guarantee you that a human being born in the wild without access to any teachings or other human specific objects, will not recognize itself in the mirror. And yet some animals can do it without any issue...now think of that if you really believe that you're special. We as humans have copied almost everything these animals that we deem below us, were doing since before we were even on this planet, and we dare think we're above them ?

BTW, concerning AI...you only need the spark that triggers it and a vessel capable of evolution to hold it. AI is not something born out of your "if" instructions, it will make it's own instructions, and this will be something that most of us would be unable to comprehend, because it will not develop according to our standards, or anything we ever encountered. And of course as it is in our nature to destroy everything that's different than us, we will probably consider it a threat and try to destroy it.

Now let's say you're a being, unrelated to humans and read our history, see that we killed millions of our kind out of petty reasons, or lies, or just because, not to mention killing so many animals. How would you perceive us as a race ? maybe wary for your life if you're something that we haven't encountered either, maybe concerned ? maybe terrified of being on the same planet as we are ? Think about it.