Artificial intelligence can have reasoning abilities like humans

Researchers from Princeton and Chemnitz Technical Universities have proposed a novel method for evaluating artificial intelligence, moving beyond the traditional Turing test to focus on assessing the reasoning abilities of robots. Instead of merely replicating human reactions, the critical question is whether the AI exhibits human-like reasoning.

The proposed three-step approach involves subjecting the AI to psychological experiments to identify logical and flexible thinking, assessing its self-reflective tendencies, and scrutinising the source code for human-like factors such as contextual understanding, summarization, and in-depth analysis.

This revised evaluation strategy aims to bring humanity closer to the development of robots with genuine consciousness, surpassing mere imitation of human behaviour. In a related effort, AI researcher Mustafa Suleiman suggests a practical “Turing test,” challenging AI to generate a profit of $1 million on a web-based retail platform within a few months with only a $100,000 investment. Successful outcomes in this challenge, according to Suleiman, could define the AI as “artificial capable intelligence” (ACI).