A big debate about the rapid expansion of artificial intelligence (AI) systems has been whether AI systems will gain cognition, the mental process of acquiring knowledge and understanding through thought, experience and senses, similar to humans. According to a new study, that is unlikely.
The research, published in the journal Science Robotics, found that AI systems will not gain brain processing similar to humans regardless of how advanced or large their neural networks or the datasets used to train them become if they are not connected to the real world through robots, according to a report by the Press Trust of India (PTI).
The research team from the University of Sheffield, UK, said that the existing AI systems such as the popular ChatGPT, use large neural networks to solve problems such as generating intelligible written text. These teach AI to process data similar to how the human brain and also learn from mistakes to enhance accuracy, the PTI report adds.
While they exhibit processes similar to the human brain, there are important differences which will prevent them from gaining intelligence similar to humans. These differences include a physical system such as the human body that directly senses and acts, according to the PTI report. Having a body makes these processes meaningful by adding a direct connection to the physical world which AI lacks, the researchers explained. This is the reason such AI systems do not have awareness of the real world.
Moreover, the human brain has multiple subsystems which are organised in a particular structure that is similar in all vertebrate animals but not in AI. The study suggests that biological intelligence developed because of this structure and this structure has used its connections to exist and evolve. This link between evolution and development is rarely considered when designing AI, according to the researchers.
"ChatGPT, and other large neural network models, are exciting developments in AI which show that really hard challenges like learning the structure of human language can be solved," said Professor Tony Prescott, Professor at the University of Sheffield, in the PTI report. However, such AI systems are unlikely to reach a stage where they can think like a human brain if designed using the same methods, Prescot added.
“It is much more likely that AI systems will develop human-like cognition if they are built with architectures that learn and improve in similar ways to how the human brain does, using its connections to the real world," he explained in a press statement.