AI achieved great success in understanding human emotions, GPT-4.5 shocking performances in Turing Test

Today’s digital age is changing rapidly and the major reason for this is Artificial Intelligence (AI). Whether it is to write a script, to make a video from a photo or analyze data – AI has now reached very close to human abilities. Recently, scientists at the University of Osaka, Japan have developed an AI model that can estimate the biological age of a human being.

In this episode, two new AI models have now appeared which are considered to be able to understand and behave like human feelings. These models are Openai’s GPT-4.5 and Meta’s LLAMA-3.1, who has passed an important turing test.

What is Turing Test?

The Turing Test is a criteria proposed by famous mathematician Alan Turing in the 1950s. Its purpose is to find out if a machine can be so intelligent that it can give conversation and emotional response like a human being.

Till now passing this test is considered a big achievement for any AI.

GPT-4.5 results are very impressive

Researchers at California University, San Diego, Cameron R. Jones and Benjamin K. Bargon found in his research that GPT-4.5 showed about 73 percent accuracy in the Turing Test.

• This means that the reactions and behavior of GPT-4.5 looked like humans up to 73 percent.

• This AI model was not only proficient in dialogue, but was also showcasing emotional understanding.

Meta’s LLAMA-3.1 performance is also notable

• The LLAMA-3.1 model of Meta remained close to human behavior with 56 percent accuracy.

• At the same time, baseline models like Eliza and GPT-40 were able to behave only like humans only 23% and 21%.

Has AI become like a human being?

Researchers say that passing the turing test does not prove that AI has been able to fully adopt human thinking and behavior.