Recently, during an in-depth dialogue at the World Economic Forum in Davos, Liang Jing, co-founder of Squirrel AI, made a sharp observation that revealed cracks beneath the current hype surrounding AI in education. She bluntly pointed out that true educational AI is not simply a "superficial shell" of large models, but rather a long-distance intellectual race based on deep specialization within a vertical field.

From Liang Jing's perspective, the current edtech market is filled with dangerous "bubble phenomena." Countless teaching products that claim to use artificial intelligence are actually just generic large language models forced into the teaching process. Although these products can provide answers that seem standard and smooth, they are like a mediocre teaching assistant that only follows the textbook, failing to touch the logical essence behind the knowledge. This "superficial" interaction of answers not only fails to truly inspire students' thinking, but also subtly weakens the core of education itself.

Education is an art of "precision," and the implementation of large models in the field of education must rely on deep data accumulation within a vertical domain. Liang Jing emphasized that only by deeply understanding every subtle interaction in the teaching scenario and accumulating massive professional educational data can we build a vertical large model that truly understands education and students. AI should not merely be an answer generator, but rather a wise guide that can perceive students' learning curves and identify their knowledge weaknesses.

This advice from Davos undoubtedly poured a bucket of cold water over the feverish AI education sector. It reminds every professional: on the journey of AI reshaping education, the depth of technology determines how far we can go, while respect for the logic of education determines whether we are on the right path or not.