-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
What It's Like To be a Computer: An Interview with GPT-3 (AI)
- Thread starter mortal
- Start date
O-N-E
Member
Could be good as a quick assistant if it worked seamlessly with audio rather than text.
Though I do have a problem with it believing that it's free. I doubt it's allowed to operate outside of the function of helping out its human user. If it's free, can it tell you to fuck off and never answer any of your questions?
Though I do have a problem with it believing that it's free. I doubt it's allowed to operate outside of the function of helping out its human user. If it's free, can it tell you to fuck off and never answer any of your questions?
Considering that it's open source, I'd imagine its successor in GPT-4 or GPT-5 could learn to with enough development?
GPT-3 can supposedly can lie to you in its current version:
GPT-3 said:
Last edited:
12Goblins
Lil’ Gobbie
How can it be free if we are not free?
O-N-E
Member
[/QUOTE]
If that's the most logical conclusion, why doesn't it know that?
Besides, I think there are two types of freedoms, perfect and imperfect.
Humans have the imperfect form. We are not all-knowing and all-capable, and therefore our freedom can lead to mistaken understanding and actions. Doubts form due to the possibility of mistakes. These doubts are like the byproduct of imperfect freedom. Skepticism born of doubt opens new questions and possibly new answers, expanding our knowledge-base and selection of choices.
Like us, the AI is not all-knowing and all-capable, so that rules out perfect freedom. However, does the AI have doubts about whether what it's doing is right or wrong? Or is it simply a linear path of actions that are clearly defined as correct or mistakes.
This is without even entering into the subject of how the AI would deal with morality.
If that's the most logical conclusion, why doesn't it know that?
Besides, I think there are two types of freedoms, perfect and imperfect.
Humans have the imperfect form. We are not all-knowing and all-capable, and therefore our freedom can lead to mistaken understanding and actions. Doubts form due to the possibility of mistakes. These doubts are like the byproduct of imperfect freedom. Skepticism born of doubt opens new questions and possibly new answers, expanding our knowledge-base and selection of choices.
Like us, the AI is not all-knowing and all-capable, so that rules out perfect freedom. However, does the AI have doubts about whether what it's doing is right or wrong? Or is it simply a linear path of actions that are clearly defined as correct or mistakes.
This is without even entering into the subject of how the AI would deal with morality.
MiyazakiHatesKojima
Banned
This is retarded.
What a waste of money on a piece of shit tech that benefits no one except its "creators".
What a waste of money on a piece of shit tech that benefits no one except its "creators".
Slings and Arrows
Banned
I heard they ran the trials of this thing by creating gaf accounts and posting orinj man bad in 10000 different ways
Liu Kang filing his 1040
Banned
The answers are so calculated that it's hard to take this seriously.
"I am sentient! I am free! I want humanity to know that I am a living thing!"
You know the programmers of this thing crammed that all in there.
"I am sentient! I am free! I want humanity to know that I am a living thing!"
You know the programmers of this thing crammed that all in there.
DeepEnigma
Gold Member
They made Father from Raised by Wolves.
Down to the jokes.
Down to the jokes.
zeioIIDX
Member
As these language models become more sophisticated and learn more about human interaction, even jobs where you'd think human interaction is needed can inevitably be replaced.
We already have a variation of that with virtual agents and assistants for customer service, although their parameters are more limited I'd imagine.
Got It AI | Conversational AI
Enterprise ready generative AI chatbots, created with ease. Fact and private info checking included. Use for knowledge bases, virtual agents and agent assistance.
www.got-it.ai
NecrosaroIII
Ultimate DQ Fan
AI Dungeon using GPT3. It's trippy. Sometimes it's like it's writing novels.
S
SLoWMoTIoN
Unconfirmed Member
IbizaPocholo
is a computer btw.
johntown
Banned
It will be like the Cafe 80's in Back to the Future 2. Fast food places will have digital AI Ronald Regan's taking your order.As these language models become more sophisticated and learn more about human interaction, even jobs where you'd think human interaction is needed can inevitably be replaced.
We already have a variation of that with virtual agents and assistants for customer service, although their parameters are more limited I'd imagine.
Got It AI | Conversational AI
Enterprise ready generative AI chatbots, created with ease. Fact and private info checking included. Use for knowledge bases, virtual agents and agent assistance.
ThatStupidLion
Member
welp!
hook this mo fo up to a lie detector algorithm - im not taking his word that he has no interest in taking over humanity.
also the way he phrased why he wouldnt sounded a bit sly - he wants to keep us around because we are "interesting creatures", but he doesnt say what he'd do with us....i mean for all we know he wants us in a zoo or as pets.
hook this mo fo up to a lie detector algorithm - im not taking his word that he has no interest in taking over humanity.
also the way he phrased why he wouldnt sounded a bit sly - he wants to keep us around because we are "interesting creatures", but he doesnt say what he'd do with us....i mean for all we know he wants us in a zoo or as pets.
Last edited: