Cultivating talent that machines can't match

How do we identify talent? For generations, standardized tests were the answer. They helped universities find promising students, employers recruit capable graduates, and individuals demonstrate academic achievement regardless of their background.

Now, artificial intelligence is shaking up those assumptions.

Students can ask AI to solve complex equations, write computer code, summarize academic papers, or draft essays in seconds. As AI gets better at performing the kinds of tasks that have traditionally been tested, educators are asking an important question: Can standardized tests still measure talent?

The answer is yes, but only if we rethink what talent is worth measuring. Knowledge still matters, as students cannot think critically about what they do not understand. But when information is readily available, education needs to evolve.

This change is already apparent in higher education. AI is increasingly helping with routine cognitive tasks such as writing reports, translating documents and analyzing data. As leading AI systems perform exceptionally well in many conventional academic assessments, educators need to ask whether those assessments are measuring genuine human capability or simply the ability to produce correct answers.

In future assessments, basic knowledge should still be evaluated, but greater emphasis should be placed on higher-order thinking. Can students think critically about contradictory evidence rather than simply memorize facts? Can they solve problems in a novel manner instead of repeating familiar procedures? Can they distinguish reliable information from compelling misinformation? Can they explain not only what decision they reached, but why and how they reached it?

Creativity, curiosity, ethical judgment, communication, collaboration and resilience are also becoming increasingly valuable. These qualities enable people to ask new questions, learn continuously, make responsible decisions, work with others and persevere when problems have no obvious solutions.

These qualities are more difficult to quantify than factual knowledge, but they are not impossible to evaluate. Assessment needs to look at learning outcomes in new ways in the AI age. Learning how to learn is one important outcome.

Imagine two students submitting equally polished essays. One simply accepts an AI-generated draft with minimal revision. The other uses AI as a sparring partner, questions its assumptions, verifies its evidence, strengthens its reasoning, and develops an original argument. Traditional assessment may see identical end products. A more sophisticated assessment looks at the thinking process that produced them. In the AI era, that process may matter more than the product itself.

In some courses, instructors might even ask students to submit the AI-generated answer alongside their own work. If a student's submission merely mirrors what AI produces, it deserves only a passing grade at best. Students should instead be rewarded for going beyond AI, questioning its assumptions, identifying its weaknesses, integrating additional evidence, and developing original insights.

Instead of banning AI altogether, more universities are teaching students how to use it responsibly while placing greater weight on classroom participation, discussion, presentations, and other forms of active engagement where students must think on their feet, defend their reasoning, respond to challenges, and build on one another's ideas. AI can assist preparation, but it cannot replace the critical thinking, intellectual curiosity, and collaborative judgment demonstrated in the classroom or in real life.

Duke Kunshan University exemplifies this shift by integrating AI into learning while placing greater emphasis on discussions, presentations, authentic projects, reflective writing and community engagement.

Every DKU undergraduate also completes a faculty-mentored Signature Work project, an original piece of research or creative work that frequently engages real-world questions or challenges. Students are assessed not only on what they know, but also on their ability to ask meaningful questions, synthesize knowledge across disciplines, conduct independent inquiry and communicate their findings effectively. These are precisely the kinds of capabilities that AI can support but cannot replace.

China has invested heavily in artificial intelligence and made AI education a national priority. It is also promoting educational reform that emphasizes creativity, interdisciplinary learning, and scientific innovation alongside strong academic foundations. As these investments continue, China can become a global leader not only in developing AI technologies but also in designing assessment systems that identify the kinds of talent needed for an intelligent economy.

As machines increasingly generate information, universities must cultivate the ability to apply knowledge responsibly in complex human situations. Ultimately, the purpose of education is not to produce good test takers. It is to develop citizens who are capable of making reasoned judgments, leading responsibly, and remaining lifelong learners.

AI is changing how we learn. It is changing how we work. It should also change how we evaluate talent. The most successful societies will not be those that produce graduates who can outperform AI on yesterday's exams. They will be those that cultivate people who know how to collaborate with AI while contributing what machines cannot: imagination, judgment, resilience, integrity and compassion.

These are the qualities that our schools should increasingly seek to identify, our universities should nurture and our examinations should reward.

The author is executive vice-chancellor, American president and distinguished professor of social science at Duke Kunshan University in China.

The views do not necessarily reflect those of China Daily.

If you have a specific expertise, or would like to share your thought about our stories, then send us your writings at opinion@chinadaily.com.cn, and comment@chinadaily.com.cn.