We are living through a race to build smarter machines. Governments are investing in artificial intelligence (AI), universities are integrating it into teaching, and businesses are competing to automate human tasks. Technological leadership is increasingly treated as a measure of national progress. But an uncomfortable question remains: what if a nation gradually sacrifices its human intelligence by becoming increasingly dependent on artificial intelligence?

The nations that dare to resist artificial intelligence today may become the nations richest in human intelligence tomorrow.

This is a deliberately provocative argument, not a call to abandon technology. It challenges the assumption that adopting more AI necessarily makes people more intelligent. A nation may become technologically powerful while its citizens gradually lose the habit of thinking independently.

Consider two students using AI. One asks it to explain a difficult concept, checks the explanation and develops an independent understanding. The other asks it to complete every assignment and solve every problem, accepting the results without examination. Both receive answers, but only one necessarily develops greater understanding. The first uses technology to support thinking; the second risks using it to avoid thinking.

Intelligence is more than obtaining information. It involves questioning assumptions, evaluating evidence, recognising contradictions and making independent judgements. These abilities develop through practice. When we transfer too much of our thinking to machines, we risk weakening the very skills education should strengthen.

Robert Sparrow and Gene Flenady (2025), in Bullshit Universities: The Future of Automated Education, raise this concern in higher education. They argue that learning involves more than acquiring facts: students must learn how to think, practise skills, observe human role models and participate in intellectual communities. Their warning is important because education is not simply the delivery of information. It is the development of minds capable of judging information independently.

Generative AI can produce convincing explanations and fluent essays, but fluency does not guarantee accuracy. Research on natural language generation identifies hallucination as a significant problem: AI systems can produce information that sounds credible but is false or unsupported (Ji et al., 2023).

The danger is particularly serious for learners who have not developed the knowledge needed to recognise errors. If students accept machine-generated answers without checking them, they may become dependent on systems they cannot adequately evaluate.

This does not mean AI has no educational value. Kasneci et al. (2023) discuss its potential to support learning, provide feedback and assist teachers, while highlighting concerns about accuracy, bias and misuse. The essential distinction is between assistance and substitution. A tool that challenges a student's reasoning can strengthen learning. A tool that routinely performs the reasoning on the student's behalf may deprive the student of opportunities to develop it.

Writing offers a clear example. When students struggle to express an idea, they often discover weaknesses in their understanding. Revising an argument, finding the right words and correcting mistakes are part of intellectual development. If AI performs the entire process, students may receive polished assignments without acquiring the abilities those assignments are supposed to demonstrate. A beautifully written assignment is not necessarily evidence of a well-developed mind.

Universities face a particular risk because AI promises efficiency: faster marking, personalised explanations and reduced workloads. Yet what happens when machines generate assignments and other automated systems evaluate them? Education could become a process in which technology produces answers and technology assesses them, while the student's actual learning becomes difficult to measure.

Sparrow and Flenady (2025) warn against treating the replacement of human teachers as an inevitable consequence of technological progress. Teachers do more than explain subjects. They demonstrate curiosity, intellectual honesty and the willingness to reconsider an argument. Students also learn through discussion, disagreement, feedback and collaboration.

Evidence supports the importance of active participation. Freeman et al. (2014), analysing 225 studies in undergraduate science, technology, engineering and mathematics, found that active learning improved examination performance compared with traditional lecturing. This reinforces a central point: meaningful education requires students to participate in learning, not merely receive information.

AI can support this process, but it should not eliminate the human relationships and intellectual effort through which education develops.

Completely rejecting AI would be neither practical nor desirable. It can help researchers analyse information, assist people with disabilities, support translation and provide learning opportunities to students who lack access to specialist teachers.

The challenge is to use AI without surrendering the responsibility to think. UNESCO's guidance on generative AI in education and research advocates a human-centred approach that protects human agency while addressing ethical, educational and social risks (Miao & Holmes, 2023).

Students should attempt problems before consulting AI, write independently before requesting feedback, and explain their reasoning rather than simply submit machine-generated answers. Universities should use discussions, practical tasks, oral examinations and supervised assessments to establish what students genuinely understand. AI literacy must include knowing when to question a system, verify its claims or work without it.

Governments should judge educational innovation by the capabilities students develop, not by the number of processes automated or the money saved.

The most advanced nation of the future may not be the one that delegates the greatest number of tasks to machines. It may be the one that preserves the greatest capacity for independent thought while using technology intelligently.

Human intelligence allows societies to question authority, recognise injustice, challenge popular assumptions and imagine alternatives. These capacities matter in science, public life and democracy. A nation that neglects them may become technologically sophisticated yet intellectually vulnerable.

The purpose of education is not simply to produce people who can obtain answers quickly. It is to develop people who can determine which questions deserve attention, which answers deserve belief and which problems require a new way of thinking.

Resisting AI does not mean refusing the machine. It means refusing to surrender the mind. The nations that dare to resist artificial intelligence today may become the nations richest in human intelligence tomorrow—not because technological ignorance is a virtue, but because wisdom lies in recognising the limits of convenience. The future will need people who can operate powerful machines. More importantly, it will need people who can think clearly when those machines are wrong, silent or absent.

The ultimate test of progress is not how much thinking we can transfer to machines, but how much human intelligence we preserve while doing so.

References

1. Freeman, S., Eddy, S. L., McDonough, M., Smith, M. K., Okoroafor, N., Jordt, H., & Wenderoth, M. P. (2014). Active learning increases student performance in science, engineering, and mathematics. Proceedings of the National Academy of Sciences, 111(23), 8410–8415. https://doi.org/10.1073/pnas.1319030111

2. Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., Ishii, E., Bang, Y. J., Madotto, A., & Fung, P. (2023). Survey of hallucination in natural language generation. ACM Computing Surveys, 55(12), Article 248. https://doi.org/10.1145/3571730

3. Kasneci, E., Seßler, K., Küchemann, S., Bannert, M., Dementieva, D., Fischer, F., et al. (2023). ChatGPT for good? On opportunities and challenges of large language models for education. Learning and Individual Differences, 103, 102274. https://doi.org/10.1016/j.lindif.2023.102274

4. Miao, F., & Holmes, W. (2023). Guidance for generative AI in education and research. UNESCO.

5. Sparrow, R., & Flenady, G. (2025). Bullshit universities: The future of automated education. AI & Society, 40, 5285–5296. https://doi.org/10.1007/s00146-025-02340-8