Artificial intelligence has just crossed a frontier that seemed impossible, for the first time in history an artificial system passed the famous classic Turing test, convincing the majority of people that it was a human being, according to a new study from the University of California in San Diego, GPT 4.5 managed to convince humans that it was human too, with a success rate of 73%, but that means that it thinks or we are simply facing a highly efficient conversation simulator.
The researchers carried out a rigorous study testing four different systems Eliza, GPT 4.0 Yamaha 3.1 and GPT 4.5. Each human participant unknowingly conversed simultaneously with another human and with an AI. After 5 minutes of dialogue, the participants had to identify which interlocutor was the human and which was the machine.
The result surprised the GPT 4.5 with specific instructions to adopt a human person convinced 73% of people, even surpassing real human participants, the LLaMa-3.1 model had a performance close to human with 56% and while older systems such as ISA and GPG4 were far behind with only around 20%.
The success of GPT 4.5 is directly related to the human strategy adopted by the AI, during the tests the GPT was instructed to simulate personal characteristics, emotions and behaviors typical of a real person, this approach guaranteed more natural and convincing dialogues, however by removing these specific instructions the performance of GPT dropped drastically 36%, demonstrating that its greatest strength is in the ability to reproduce emotionally consistent and human dialogues, and not necessarily a deep understanding with real science.
This historic achievement reopens an important philosophical debate. The Turing test really measures artificial intelligence or only shows ease of believing in well-structured dialogues. Some critics argue that the brief and superficial interaction facilitates the illusion since the AI manages to manipulate language in an extremely convincing way even without possessing true consciousness or understanding.
With the ability to communicate interchangeably like humans, systems like GPT 4.0 have the potential to transform various areas: customer service, education, emotional support, even political support, however, this raises important ethical questions about how we should use and regulate these advanced technologies. In addition, it opens the possibility of deeper integration between humans and already in daily life, generating new opportunities and unprecedented challenges in the way we interact with intelligent machines.
GPT 4.5 does not just represent a technological advance, it marks a turning point in the history of artificial intelligence. For the first time we have concrete evidence that a machine can pass itself off as human, and it is surprisingly effective, but that means that it already really thinks or we are facing a new type of illusion. I want to know what you think. Does passing the Turing test make a machine really intelligent or has it just learned to deceive us better?
Thank you for visiting my blog. If you like posts about #science, #planet, #politics, #rights#crypto, #traveling and discovering secrets and beauties of the #universe, feel free to Follow me as these are the topics I write about the most. Have a wonderful day and stay on this great platform :) :)
! The truth will set us free and science is the one that is closest to the truth!
Grazie per aver visitato il mio blog. Se ti piacciono i post su #scienza, #pianeta, #politica, #diritti#cripto, #viaggi e scoperta dei segreti e delle bellezze dell'#universo, sentiti libero di seguirmi perché questi sono gli argomenti di cui scrivo di più. Trascorri una splendida giornata e resta su questa fantastica piattaforma :) :)
! La verità ci renderà liberi e la scienza è quella che più si avvicina alla verità!