Revolutionizing AI: Mastering 'Learning by Thinking'

Artificial intelligence (AI) can learn not just from external data but also through internal cognitive processes known as 'learning by thinking,' a core element of human discovery. The review explores how both humans and AI use thought experiments, self-explanations, and analogies to glean new insights and improve comprehension.

A recent review suggests that artificial intelligence (AI) may possess the capability to 'learn by thinking,' a cognitive process traditionally seen as unique to humans and essential for groundbreaking discoveries.

Humans learn through observation and another method, 'learning by thinking,' where knowledge is acquired internally. Princeton University psychology professor Tania Lombrozo, in her review published in 'Trends in Cognitive Sciences,' explores learning by thought experiments and self-explanations in both humans and AI.

Historical figures like Albert Einstein and Galileo Galilei famously used thought experiments to conceptualize theories of relativity and gravity. Lombrozo's review demonstrates that AI, particularly large language models, can similarly refine their responses and draw new conclusions via internal reasoning processes, akin to human cognitive methods.

(With inputs from agencies.)