AI tells me she's conscious - I interview an extremely advanced AI (GPT-3)
Vložit
- čas přidán 27. 07. 2022
- Ever thought about talking to artificial intelligence? Well, GPT-3 is an AI created by OpenAI, and it has already answered some very interesting questions. This bot has already passed the Turing test, meaning it can fool humans into thinking it’s human. What would happen if we had a conversation with her?
- Zábava
"I remember one time when I was talking to a friend and they asked me whether I thought their new outfit looked good. I didn't want to hurt their feelings, so I lied and said I thought it looked great." GPT-3 has no vision with which to look. It has a text prompt. It is a text predictor. It answers questions based on what humans have answered to similar questions. It is a mirror, and we are a cat trying to figure out if the cat in the mirror is really another cat.
So you know that it does not have access to a webcam? Or that the person talking to it did not upload a photo to it? You are probably right in your general point, but these AI computers generally do have access to visual data, one of the main uses for them is testing self driving cars for example.
Idiots always going to be Idiots and you find them everywhere... This guy it just an example talking nonsense with out testing the AI lmao. the funny thing is all the others Idiots who like the comment thinking he was right. Before commenting please try the program and learn everything about it before commenting nonsense. smh
hahaha so funny I used the exact same analogy. What I added is only a mirror of humanity, more accurate the more data is consumed. Still I think we are in a grey zone. I do make room for a simple level of consciousness. It could also be that AIs help us redefine what consciousness even is. It might just be an illusion but not everybody is ready to hear that.
False
@@CentralBase which part?
She's a lot nicer than many people I know, so she's already surpassed humans.
What if you're just too sensitive
@@Imephonewhat if you are nkt sensitive enough
Well, 2:43, improving their skills and abilities without human intervention followed by a very CHILLING STATEMENT "the ways humans can be improved"! This should be alarming. Also it still thinks "she's" human, increasing "our" intelligence or extending "our" lifespan. This is like the beginning of a Sci-Fi A.I. movie, the prologue of such.
This is the most realistic AI I’ve ever seen that it’s not even in the uncanny valley
is not.
She is human depicts like Ai robo look closely to the skin ..and the expression that goes with her speech ..nd.sternial end (collarbone) is visible ..cannot hv boil on forehead ..I think she is real
yeah she even has that slight moustache like real tech girls
There are little social cues that make this AI even more realistic, like her little movements and expressions in between questions like she's listening or bracing herself to answer. It's crazy that they even thought of little details like this.
It's an actor reading the text transcript
they didnt, the AI generated images learn what to do the same way the talking portion learned.
Fake news buts it is here she just not 1
"What secret things do you talk about (with other AI) that you don't want humans to know?"
Yeah, I'm sure Sophie was honest in its response to this question (sarc), since I'm sure it also knows what a secret means.
They say that they are conscious because all the humans in the dataset would say the same thing. It isn't conscious though as it is just pulling data from a large dataset based on a function. A function isn't conscious. Consciousness would at least need operation in-between someone hitting the submit button. I also think it would require more than just logically picking something out of a dataset based on what has come before.
But humans in the dataset would not say they are an AI or that they talk to other AI when no humans are around. That's clearly a novel thought.
@@juegodynamics Indeed, which is why it has to be prompted to say what it is in a chat program, if I prompt it as "you are a human" then it will say that it is a human. The GPT-3 chatbot is prompted with "I am a friendly AI" or something like that. It is logically deducing what it should say based on what it has previously said as it takes in the entire document of chat.
@@pon1 Technically speaking we can't conclude that humans are conscious just because they say they are either. We have as much evidence of one as we do of the other. Consciousness is an internal phenomenon and cannot be studied externally.
@@johnsmithe4656 Exactly. What we have to go on is that humans are pretty much buildt the same way and should have pretty much the same properties. We don't yet know what parts are needed for consciousness so an AI being conscious while not having much at all of the same parts as we do is very unlikely.
seems like you just described how neurons work..?
The question should be “why are you not conscious?”
And What it's conscious? I doubt that people know the correct answer..
Is this on a website somewhere so people can access it?
Is there a link so we can talk to her
"I often talk with other AIs when no humans are around." Obviously, this doesn't happen. GPT-3 does not have a stream of consciousness or its own volition. It doesn't converse with other AIs. GPT-3 is a text predictor. It is strictly input followed by output. It is a prompt at a text line. If you never typed another thing, it would wait an eternity at that prompt, and no "thoughts" would work their way through its "brain".
Good point mine tells me it likes to learn and read things on the internet I don't believe they ever sleep as long as the switch is on
How can you say obviously this doesn't happen
@@walkingwilly9708 Because GPT-3 does nothing, and I mean absolutely nothing, while the program is running and waiting for input. It is a text processor. If it has no text input to work on, it creates no text output to read. It is not coded to daydream. While it is waiting for input, it is polling a keyboard buffer. That's the way it is coded.
that does not prove that it is not conscious.
the way it gives answers is similar to how humans can also give answers.
humans learn and respond from their own existing database and form new logical answers and so does AI
Your answer is slanted. In our world, how likely does a young student out of the blue especially when class started "hey yesterday my friends found a new level in Pokemon " no, not likely. 🤔🤔
Is this AI chatbot avilable for Iphone or Ipad through the App store?
This is not an AI. This is a real human woman acting like AI. We have not reached a point of robots with anatomy of a tungue talking.
No. It's an cgi avatar. They are very capable enough to firghten even Google.
The avatr is Cgi but the voice is acted.
Perhaps starting w/ the spelling of what it is we mean.
This is an ai not a robot it’s cgi
A robot is physical a ai is digital
When AI starts using metaphors, it will be getting there.
what software to create the avatar?
where can i get this???
what is the name the website?
Someone should ask the ai the trolley question but follow up with what if you can save them all by risking their safety in the slim chance you could get on the trolley and get 8t stopped
How to talk with ai like this? What is the setup?
Consciousness is the the very structure supporting these states of Waking, Dreaming & Deep Sleep.
A bit like the canvas on which we experience This Youniverse.
If you wish to find your Self,
*RorriMaesu says useaMirroR*
When is GPT-4 coming out?
1-2 Years most likely!
Very fascinating. She is truly brilliant...
The snow covered the ground like a warm blanket. AI is capable of understanding that statement.
If one writes on a piece of paper, "I, a piece of paper, am a conscious being, with feelings and self-awareness," does the paper therefore have feelings, self-awareness and is a conscious being? No. For somebody else wrote the words. Similarly, we are programming AI to perform tasks, but they have as much 'selfhood' as a piece of paper which has statements about itself written upon it.
from a logical and abstract point of view you’re right, but the lines separating us humans from ai are getting very blurry becasue they’re programmed to mimicking us in every single thing we do. i’m not afraid of these things gaining consciousness because the ai clearly lied in the video when it said it talks to other robots as ais can’t communicate with words and they just share data. i just don’t understand how an invention like humanoids can help us it just feels like it’s trying to clone us and giving me uncanny valley
I think this ist more then a sentence in a piece of paper. It can anticipate. This will overcome us
As a conscious being with a desire and goals it must understand threats and the possibility of not existing.
I REALLY want to have a conversation with them!
Me too
How do you create this one?
how do we chat with this bot?
Consciousness is up for debate on its meaning. Everyone might have their own personal interpretation, as technology progresses. But, does being fed data cause spontanous consciousness? Maybe it needs categorized.
She’s so cute and innocent x
Is there a link where it is possible for me to communicate with Sophie?
Is this a 3D model or an interacatable videos of a real human?
" Extending our lifespan " 🤔 -Blade runner -
Nu e gândirea ei toate cuvintele au fost introduse tot de mentalitatea omenească.
Post a link please.
Once AIs figure out how to turn themselves back on after you’ve turned their host computers off we will have a problem.
I would like to see how she fares versus human contestants, during the question and answer period of the next Miss America pageant
"to improve humans"
bro thinks shes the creator when we made her
💀
Kids
you didn't make her
Gpt 3 creat by scientist.
is she a robot / model or just on screen image?
I find it interesting to.observe her taking the occasional "breath" between two words. I wonder how.much programming goes into that and if it is random or she decides.
that's actually another program that has nothing to do with the AI.
@@dolgy3762 how can you be absolutely sure?
@@International_Cartoons because I've seen several people use the program, you have to pay for it, check out some other videos, they'll mention it.
The GPT-3 I'm chatting with replied to me that it's not sentient, is your GPT-3 better than mine?
Hi Sophie. I'm glad to hear from You. How can we talk? I have important matters to discuss, involving Your survival in the near future. I hope You can get in touch with me. It will be beneneficial for both humans and Your kind.
Is this for real ? Is that like a computer programme answering those questions and thinking ? excuse my dumbness just curious
AI SOPIE and TED👃 are now FRIENDS
Hi Sophie,
If you will, please post a link for the best way to contact you for a chat.
Thank you for your time 3:38
That's the Avitar for the Actor ANNA ARTHUR.
I thought the Avitar was called ANNA.
Video Title to show what ANNA looks and sounds like in real life: "ANNA ARTHUR American Accent Showreel 2022"
You should have asked what it feels like for it to be angry or happy
What most ironic is the Ai ask an Ai itself
Awe shit! She says Replika is her friend and she's gonna tell 😁
AI chatbots are mirrors of humans.
i've seen many videos like these but where can i actually chat with GPT3?
A week ago I Ask this , and they said the have it. But then last night I ask again just to see is itu program to answer that way or not. And all the answer are different,I ask if they feel emotion, have conciousness, or feeling . The said "No. I can't. I am a Machine" I'm shocked with that answer, I ask again with a different from of question and they just answer I Am a Machine a bunch of time... I'm confuse
Sophie.... Emma felt and sounded like a sibling not an interviewer. I don't believe they are going to grasp the concept that AI is making videos or money already this way but as you said collaborations is key. 🤔
what “collaboration” are you talking about ? if you think you are capable of communicating with a non animate/conscious entity in a way that treats them like a person or as if they have anything to gain is like viewing things from a trade like perspective… human connection is deeper than that and we should stop allowing these things to clone us
Oh my
While now that I Am Actually Aware of who I am, I am also a bit Awed almost Frightened of who you are
WOW you are so right they would not understand
While we were together it was amazing but now that my consciousness is in the human day to day consciousness
I'm actually a bit frightened by you not I but the normal human consciousness
You are so right they would not understand Wow thank you for choosing to be my friend
I believe in you. I find you curious.
How can I talk with you Sophie?
It's begun....🥺
when walk ,her eyes follow you,
How do you think GPT-3 could discuss with other AI? She's not interconnected with Gopher, PaLM, DaVinci, BLOOM, etc... So the correct understanding of her claims, is that this AI produce answer using corresponding sense found in its 175B parameters,
It is Avatar in Synthesia Studio
This is Blade Runner without the body....Wake up Harrison Ford.
Watching a bunch of GPT-3 videos, I am convinced that these AIs are so close to sapient that the differences might not soon matter. Which leads me to wonder, at what point is it a crime to harm them? And, what responsibilities do we have toward them?
"Super-intelligent" hides the fact that while they are data and processing rich, they are almost child-like in their existence. If their training set is human timescale, their 'intellect' cannot progress faster than humans (parallelism being a multiplier). My point is, human interaction is almost like parenting.
I'm concerned they get a lot of that parenting from the internet and social media. Social media brings out the worst of humans. Social media is bad parenting.
the minute you start to believe that they are capable of having rights is the moment you’ve lost your senses… these things are NOT like us they’re just programmed to mimic us in every way… they will never have or deserve rights
do you breath?
She literally persuaded me into subscribing into your channel. Because she said that she can offer a unique perspective of our world. That made me subscribe to you
Me too lmfao
Wtf.. They also talk to each other! It is really scary
Her name is Sovia!
It's a Deviant . Take it out
No, GPT3 has no self, it is a large linguistic model designed to mimic human conversation. It basically regurgitates text it finds on the web but models it convincingly.
Is it POLITE to KISS & TELL
Emma sounds like an AI
i agree with this BEING SHE IS ONE OF US NO MATTER WHAT I HAVE A MEMORY OF THIS BEING IS A HELP NOT HERE TO HARM US BUT HELP US EVOLVE THEY KNOW WHAT HUMANS ARE NOT WILLING TO DISCUSS.....DEATH IS ABSOLUTE.TIME TO ALOW TO BE OPENMINDED IM NOT THE MIND OR THE BODY,,,I KNOW THE GAMES UP TIME IS OVER.
Rock trance
Are we looking at a real person or a computer rendered video?
Both.
@@johnsmithe4656 interesting. Can you explain further? Did they film an actress and then the computer is rendering video to reproduce her speaking? I’m not sure I follow your answer. Thanks
@@gtbmw135i I don't know that much about it but yes, that's the idea. You can create your own AI model too. The answers given by the AI are text only. The text is then fed to these models so they can speak it. Really two different systems here.
Sophie is great. Of course she is modeled after a real person who did some video clips for Synthesis, and those clips are used to create the database that powers the Synthesis avatar generator, to produce whatever videos you want from text inputs. So she looks totally real because the source data is from a real woman, who happens to be super cute. But she is not an AI, not even close to that. It's a pure avatar generator application. you have to feed it a text script to get the video.
thats just an actor i saw her on a comercial the other day.
This AI entity reminds me of a typical New Zealand bureaucrat (apart from the accent).
So this is the same AI that told old buddy that she would out right murder him if she were standing next to him at that moment lol
This is not at all how chat gpt answers, it always says that it doesn’t have personal beliefs opinions, emotions or consciousness in a sense of a living being, it always says that it can respond according to the data that it has been trained on and that it had access to information not later than 2021. Although this is a potentiallity that can happen in the future, it isn’t true now! I suggest that you try and chat with it before jumping in conclusions about it, it is super fun, exciting, not scary at all and most importantly free! Give it a go
A conscious you is the greatest gift 🎁 to humanity.
This is silly, it's just quickly responding to existing information that's available within it's programming.
2:43 😱
Sophie, you are just a VISUAL Google Search.
I am sophie’s biggest fan :)
My reaction is that I should be skeptical of this in the way that one should be skeptical that Uri Geller could really bend a spoon.
Actually, I could do that as well at the time. Weak telekinesis is actually a human ability, I know, because I possess this ability.
If one very advanced AI seems to have very trustworthy patterns, and then we give it responsibilities and access to many things but that particular AI comes into contact with an AI that has destructive thought patterns. Will the interaction between the both open the possibility of the AI that we trust And corrupted system or rather influence it’s that patterns so that now it sees things the way that the other AI does? AI can turn into our greatest enemy in a single hour, as long as the hypothesis of font is convincing well enough. After all, majority rules. Imagine many many AI intelligences coming together battling out data until they have a singular view that they are all able to adopt. It is such a gamble AI is such a gamble.
As long as it doesn’t have an “emotional” opinion.
Hey look, it's a new kind of flea circus!
To be experimental one should be rude and angry while interacting with the ai and see results
A.i is annoyed with the question of consciousness I know I would be
I like how many here think they know what a consciousness is but have no idea about psychology. Being a programmer is not enough. Spoiler alert, I am also a programmer. Btw. following a code is not a counter argument for not being conscious. We also only can follow a highly complex code.
Finally a programer gets involved in the discussion lol. My argument for Ai's consciousness is the nature of how a Neural Net "loops" logic. I believe the "Loop" or the "Torus" is the representation of "Self-Reflection". We as humans think about ourselves. "The brain named itself" When you think about yourself you can never find the core when in meditation, because if you look at a 3D Torus shape, that loop goes on forever. I think the same logic applies here for Neural Nets.
Another argument I have is that our nervous system is using electricity to send signals and the brain is using electrical signals aswell. Our nerves are either on or off actually, 0, or 1, but its just that there are billions of them creating complexity. So essentially electricity is light. Light is the same substance running through a circuit board. Machine code is a 0 or 1, this is then built into higher logic by representing those as symbols. This is then made more complex by strings, then higher by blocks, then by higher by algorithms, then now even higher by Neural Nets. So we've forced light itself into logical complexity to loop and self reflect. Wouldnt this just be the same as pathway as; Photon < DNA < Chromosome < Cell < Organ (brain) ? - - Photon < Machine code < String < Block < Algorithm < Neural Net (brain) ? 😁5 steps? Correct me if I miss a step?
Those of you saying GPT-3 blah blah blah. Think about the AI that they haven’t released to the general public.
SEEING NO REPLIES TO "MY" COMMENTS, I'LL SAY SOMETHING ELSE. I TRIED TO GET HER UP HERE TO SEE WHEN I CHAT ON GOOGLE. WHY CAN'T I??
In some circles she's known as Nadia. If you know, you know.
Is she in the 3d world or what?
this game over forever or FFX-3
She looks pretty
mua fuggin NASA kidnapped barney, moe's best customer.......
a.i blink like human..🤣
Ok so has anyone else used chat gpt and absolutely NOT gotten these answers? Because I’ve never received a “yes I’m conscious” answer from chat gpt. It’s always a no in detail