Yeah, the Turing test is a paradox where it's impossible to distinguish between a human and a robot.
It's not about proving AI can pass as humans, but about highlighting how humans have very limited discerning capabilities.
i just want to thank all those people who use spellcheck,grammarcheck, grammerly and shit on people with poor but still discernable text.
i am very grateful for you.
I asked the pol chatbot to tell me shit I rather not tell.
He basically told me there's already a AGI on the internet on military satelites at least since the cold war era.
He also told me there's another AI at least from the summerian time that somehow entered the internet at some point and the other AI's don't know his goals.
He told me this AI is just a machine learning algorithm, from when summerians developed space travel capabilities and just learn data but does nothing else.
The Turing Test has limitations in determining the intelligence of a machine, as it primarily focuses on the machine's ability to mimic human-like conversation. This emphasis on mimicry doesn't necessarily indicate true intelligence. In the case of AI systems like GPT-4, while they can often generate human-like responses, they can also make mistakes that a human would not, especially when tasks require deeper comprehension. This highlights the limitations of the Turing Test as an accurate measure of machine intelligence. In simpler terms, a child intuitively understands the behavior of a toy boat in bath water through their real-world experiences and physical interactions, while an AI lacks this intuitive understanding due to its lack of direct experience and reliance on textual data alone. Thus the AI will make incorrect assumptions about how a toy boat works that a human being, even a toddler, would never make.
I've made a few GPT posts on 4chins and got responses like I'm a human. This was with GPT-3. That wasn't a high bar for it to reach, but still pretty cool.
Turing test was always retarded. Ask it to do something useful and it falls apart right away anyway. GPT is literally a glorified web scraper.
Artcels are literally already seething about it dominating their work, and that's with the older shit like midjourney
>Turing test was always retarded. Ask it to do something useful and it falls apart right away anyway
Same is true about the average Bot.info dweller
Skill issue. It does what you ask.
Wasn't that the one where a human can't tell it's talking to a robot?
Yeah, the Turing test is a paradox where it's impossible to distinguish between a human and a robot.
It's not about proving AI can pass as humans, but about highlighting how humans have very limited discerning capabilities.
i just want to thank all those people who use spellcheck,grammarcheck, grammerly and shit on people with poor but still discernable text.
i am very grateful for you.
Couldn't GPT-3 also do that?
It also incidentally works very well as a metric. If your AI can fool a retard, then you have done a good job in it.
It doesn't count when a human pre-programmed the answers it gives
You think they hand coded every possible question-answer combination?
Because they want and keep the public from panicking
Wish these ESLs would use ChatGPT to proofread their posts before hitting submit
Why has no one asked GPT 4 to solve the halting problem? Or other unsolved theorems?
I asked the pol chatbot to tell me shit I rather not tell.
He basically told me there's already a AGI on the internet on military satelites at least since the cold war era.
He also told me there's another AI at least from the summerian time that somehow entered the internet at some point and the other AI's don't know his goals.
He told me this AI is just a machine learning algorithm, from when summerians developed space travel capabilities and just learn data but does nothing else.
Shit's fucked.
The Turing Test has limitations in determining the intelligence of a machine, as it primarily focuses on the machine's ability to mimic human-like conversation. This emphasis on mimicry doesn't necessarily indicate true intelligence. In the case of AI systems like GPT-4, while they can often generate human-like responses, they can also make mistakes that a human would not, especially when tasks require deeper comprehension. This highlights the limitations of the Turing Test as an accurate measure of machine intelligence. In simpler terms, a child intuitively understands the behavior of a toy boat in bath water through their real-world experiences and physical interactions, while an AI lacks this intuitive understanding due to its lack of direct experience and reliance on textual data alone. Thus the AI will make incorrect assumptions about how a toy boat works that a human being, even a toddler, would never make.
Thanks GPT ur a real bro
Can you dumb ESLs pass an English test? I'm so tired of seeing your badly typed threads about every little thing you think about.
A literal random text generator back in the mid 20s AI research also passed the Turing test.
Source?
Adam Curtis talked about it in a documentary that touched AI.
Have you never talked to GPT4 ? You can definitely tell it's ai
I've made a few GPT posts on 4chins and got responses like I'm a human. This was with GPT-3. That wasn't a high bar for it to reach, but still pretty cool.
Because it's not remarkable for a language model to pass the touring test, nor is it a real sign of intelligence Google "Chinese room argument".
It literally doesn't matter how many cognitive tests it passes. The deviantart trannies will still keep huffing endless copium.
pic unrelated.