Current artificial intelligence systems predominantly rely on extrinsiclearning mechanisms, with gradient descent and its variants serving asthe primary means of model optimization. This approach treats learning asa distinct, external process separate from cognition. However, naturalintelligent systems, such as the human brain, display intrinsic learningwhere learning and cognition are inseparable, integrated processes.We argue for a shift of focus toward intrinsic learning in AIsystems, moving away from the heavy reliance on extrinsic optimization. Wehighlight the limitations of current AI methods, including their extremesample inefficiency and dependence on vast amounts of human-generateddata. By examining the shortcomings of current scaling approaches andproposing alternative pathways, we emphasize that genuine advancements inartificial general intelligence require systems that learn and adaptintrinsically. We encourage renewed attention of AI architecturesthat embed learning within the dynamics of the system itself, drawinginspiration from natural intelligence to foster more robust, efficient,and adaptive AI.