OpenAI’s GPT-4 language model has recently passed the Turing test, which evaluates whether a machine can imitate human behavior. This achievement highlights the advanced capabilities of artificial intelligence models and raises questions about whether GPT-4 represents true intelligence or merely a sophisticated imitation. A study conducted at the University of California in San Diego showed that more than half of the participants could not distinguish between GPT-4 and a real person.
The study also revealed that human-like AI could disrupt the economy and society by causing distrust in real interactions. Despite passing the Turing test, experts have different opinions on whether GPT-4 can be considered truly intelligent. Giovanni Geraldo Gomes, director of AI at Stefanini Latam, views the achievement positively, while Deisy Acosta Ticse, a professor of Information Systems Engineering, points out that GPT-4’s performance does not represent true intelligence but good intelligent behavior.
Acosta emphasizes the need for AI to develop towards Artificial General Intelligence (AGI), which would allow machines to understand, learn, and adapt like humans. Both experts agree on the necessity of new methodologies to evaluate AI more comprehensively and accurately. They stress the importance of metrics that cover a range of cognitive abilities to evaluate AI beyond just conversational skills.
The experts also highlight the ethical considerations surrounding the development of AI, emphasizing the need for responsible management to ensure that AI benefits society. They underscore the importance of ethics in guiding the development and use of AI to ensure its benefits while minimizing potential risks. In conclusion, they emphasize the need for human intervention to ensure that AI is developed and used in a way that is ethical and beneficial for society.