According to scientists, Chat GPT is the first robotic chatbot to have successfully passed. The Turing Test, i.e., based on human-like intelligence. Turing, a computer expert, proposed this test, in which he said an AI invention could only be declared intelligent. If no person can differentiate between the machine and themselves when talking to it.
Researchers from California University have claimed that humans find it very difficult to recognize ChatGPT-4 through written language.
What is the Turing Test (ChatGPT )?
This examination was founded by British specialist Alan Turing in 1950. He predicted that computers would one day possess abilities similar to human beings. Thus, he devised an assessment that could show whether these devices had achieved such levels of intellect.
A person participates in written conversations with humans and computers, trying to tell. Who is who—this is how the game goes during a Turing Test. If they fail to distinguish between artificial and natural intelligence, we may say our creation has exceeded human capabilities.
Chat GPT Recent Experiment
In line with this experiment, 500 volunteers were recruited by American researchers who conversed with four agents, namely three AI bots plus one human. The AI chatbots used included ChatGPT-4 and ChatGPT-3.5 alongside Eliza (a chat program from the 1960s).
All participants were put into an online chat room to determine whether their partner(s) was/were real people or just machines pretending to be so. After spending five minutes chatting, each volunteer left the virtual space and asked whether they could tell if their interlocutor(s) had been human.
Findings
Former findings showed that the Eliza program fooled 22% of people, while ChatGPT-3.5 managed to deceive half. Nevertheless, ChatGPT-4 proved most effective, as over half (54%) of all respondents identified it as a human being, thus making this robot more capable of passing the Turing Test than any previous version.
Even though 54% may not look too high, it should be noted only 67% out of those 500 volunteers managed to recognize a fellow Homo sapiens while talking with them.
According to these findings, scientists say AI systems can easily pass the two-player Turing Test.
summary (Chat GPT)
Until they are published in a scientific journal and peer-reviewed by other experts, we cannot regard these results as final, but they seem to suggest that ChatGPT-4 could pass Alan Turing’s test. The researchers said that it demonstrates intelligence comparable to humans.
