“Ya-Qin Zhang’s Agent Hospital digital environment demonstrates the potential for artificial intelligence to speed up medical training by using simulations (Nature 645, 852–855; 2025).
Universities could similarly harness AI to help students learn, by tackling complete, meaningful problems as an alternative to formal instruction.
Infants acquire language through immersion and correction, not grammar drills.
Mathematicians often grasp solutions intuitively before constructing formal proofs.
At Tsinghua University, we have built AI systems that let students immerse themselves in specific disciplines, from environmental science to industrial engineering. We observed that students who engaged directly with these systems developed deeper understanding than did those who took conventional lectures. They cultivated intuition first; formal theory then became a tool to explain what they already tacitly knew.
This sequence matters. Linear instruction privileges problems we can already articulate, producing skilled verifiers but poor problem-finders. AI can make apprenticeships more scalable by enabling unlimited personalized practice, diagnosing misconceptions instantly and delivering instruction when students encounter obstacles. This frees human teachers for mentorship, ethical reasoning and coaching.
The question for universities isn’t how can AI accelerate content delivery but how can AI help us restructure learning. That shift cultivates innovators.” [1]
This why we, the West, don’t have rare earths for magnets needed in contemporary technology. Innovation is important. We talk about it. We don’t do it.
1. Nature 646, 1060 (2025) By Shuaiguo Wang
Komentarų nėra:
Rašyti komentarą