The author, a developer, notes that AI models have significantly lowered the barrier to obtaining information. Unlike traditional forums where users had to wait for answers and navigate social friction, AI provides immediate, non-judgmental responses to diverse inquiries ranging from programming to immunology.

The author shares personal experience with developing the EVRTCK codec. Despite AI suggestions to use existing industrial solutions, the author continued with custom development to meet specific requirements. This highlights the tension between rational, efficient AI advice and the value of independent exploration.

During experiments with an artificial neural organism called ZERO, the author observed how easily one can be misled by AI-generated validation. While AI can assist in testing hypotheses, it also risks becoming a tool that reinforces a user's existing biases rather than challenging them.

The author argues that AI creates an illusion of understanding by filling knowledge gaps and providing analogies. However, true learning is tested when a person must apply knowledge without AI assistance. The distinction lies between possessing a result and understanding the underlying principles.