Gen Z meets genAI

Which AI model is best at reviewing literature? Which one is better for coding? Which produces the most natural translations? For graduate students, debating the comparative merits of different AI tools has become a daily conversation. What we don’t talk about enough is the effect these AI tools are having on our own ability to think, learn and reason. As a gen-Z international graduate student conducting humanities and social science research in Canada, in a language that is not my first, I have begun to notice a contradiction: For multilingual students, AI is both an equalizer and an obstacle to building foundations.

AI can reduce inequalities rooted in language and cultural capital. But it does more than help us find information or carry out instructions we have already formulated. It enters at the moment when an idea has not yet taken shape. It helps us name a problem, organize an argument, repair gaps in logic and determine the form in which our thinking will ultimately appear.

My generation is among the first to incorporate generative AI into everyday cognitive work while our academic abilities are still developing. University policies are still struggling to catch up with the technology, while the boundaries that define academic responsibility and separate legitimate assistance from cognitive outsourcing remain unsettled. Amid this uncertainty, our everyday practices are already producing a first draft of what it will mean for future students to learn with AI.

AI can make a text mature, but what about its author?

In the past, turning a vague idea into a coherent argument required a student to move personally through a series of difficult and often slow processes: defining the boundaries of a concept; deciding how an argument should unfold; discovering contradictions; anticipating objections; and repeatedly revising the text. These processes may appear to serve only the production of an essay. But they also shape the scholar.

In the humanities and social sciences, the purpose of academic writing is not simply to produce a polished final text. Through repeated attempts at expression, failure and revision, a person gradually learns how to think, how to judge evidence and how to accept responsibility for a claim.

Generative AI disrupts this traditional logic of learning. It allows a text to mature faster than its author. An essay may already possess a clear structure, fluent prose and an apparently complete argument, while the person whose name appears at the top has not developed the corresponding abilities to express, evaluate or defend the ideas it contains.

This is one of my deepest concerns about AI. It may help students complete assignments more quickly, but it may also help us bypass the difficulties through which learning once occurred. Once we grow accustomed to these shortcuts, we may become less willing to sustain the intellectual effort that learning requires. 

I encountered a similar problem years ago when I was learning traditional Chinese painting. While drawing the rows of tiles on the roof of an ancient building, I asked my teacher whether I could use a ruler to make the lines straight and evenly spaced. What I remember is that she did not approve of using a ruler — or, more broadly, of depending too readily on any external aids. Her response reminded me of an idea from an ancient Taoist text, the Zhuangzi: only by being “dependent on nothing” (无所待) could one “roam without bounds” (游无穷). To me, that was the mark of genuine mastery.

Today, AI resembles a ruler of incomparably greater complexity. It does not simply help me draw a straight line. It can sometimes present me with an entire plan: what I should draw, how I should draw it and how I might repair the work when something goes wrong.

AI as a cultural and linguistic interpreter

Yet I cannot simply oppose the use of AI.

I once discussed AI with a fellow student who, like me, is not a native English speaker. She told me that while living in Canada, she often relied on AI to explain cultural differences and unfamiliar situations: how to navigate an everyday purchase, how to interpret a social convention or what rules might apply in a particular circumstance.

When we began talking about how much information these systems might retain or infer about us, she shrugged with a mixture of resignation and amusement. “Honestly,” she said. “I think AI knows me better than my parents.” 

The remark was a joke, but it revealed something real. For international students, AI is not merely a productivity tool. It is becoming an interpretive system that is available at any moment. In this sense, it helps compensate for differences in language, cultural background and social experience. That is also why it so easily becomes the first place we turn to.

The dependence becomes even more pronounced in academic writing.

Academic discourse has never been solely a competition among ideas. It is also a competition conducted through language. Multilingual students may have arguments of equal value to those proposed by native speakers, but they must spend far more time managing grammar, vocabulary, collocation, tone and disciplinary jargon.

Too often, linguistic ability obscures the quality of the thinking beneath it. An idea may be no less valuable than anyone else’s, yet appear less intellectually mature because it has not been expressed in idiomatic English or in a form recognizable as proper academic prose. As a result, it may struggle to enter into serious conversation with other ideas.

From this perspective, AI has genuine egalitarian potential. It can help multilingual writers cross linguistic barriers and prevent our thinking from being so easily suppressed because of the form in which it is expressed. It can allow more people from different linguistic and educational backgrounds to participate in an academic system still largely dominated by English.

But the same tool creates a second risk. For a professor who has already developed mature scholarly judgment, using AI to improve a sentence may primarily be a matter of efficiency. For a master’s student who is still learning how to become a researcher, the same tool may impede the development of the underlying ability.

Is AI helping me express a thought I have already formed, or is it completing the formation of that thought on my behalf? The real question, then, is which forms of cognitive labour students are handing over to AI.

When AI becomes an extension of ourselves, what exactly are we trusting?

The Canadian philosopher Marshall McLuhan famously described media as extensions of human beings. AI is a technological tool, but it is also becoming an extension of its users.

Following this logic, perhaps dependence on AI should not trouble us unduly. Electricity, calculators and search engines have all become indispensable parts of modern life. Future human beings may no more be able to abandon AI than people today could comfortably abandon electricity.

But AI is not entirely comparable to electricity. While electricity is not politically neutral, its provision is generally treated as regulated infrastructure. The dominant AI systems on which students increasingly depend are different in one crucial respect: they are proprietary commercial platforms, continuously modified by companies whose training data, design decisions and long-term incentives remain largely inaccessible to ordinary users.

Are we trusting a technology we adequately understand, or an opaque system whose internal operations remain largely inaccessible to us? Are we trusting something resembling stable public infrastructure, or commercial systems developed, controlled and continuously modified by a small number of powerful technology companies?

Ordinary users cannot independently examine all the data used to train a model. Models are updated. Platform rules change. Features may be restricted, monetized or withdrawn. The data and labour of users may themselves become resources in the further development of these systems. Dependence on AI is therefore not merely a question of individual study habits. It is also a question of technological power.

When the electricity goes out, we can light a candle and return, temporarily, to an older method of illumination. But when AI becomes unavailable, unreliable or incompatible with our values, can students who have never seriously practised gathering literature, constructing arguments and writing independently take back responsibility for their own thinking?

Our generation is not only using AI. We are helping write its rules

Gen Z occupies a distinctive position. Many of us came of age academically during and after the COVID-19 pandemic, just as generative AI began its rapid ascent. But our institutions were not ready. We are using it while the technology, the institutions around it and the standards governing it are all still taking shape.

That means we should not see ourselves only as a generation being transformed by AI. We are also part of the first generation of learners participating in the creation of these norms. The habits we establish, the boundaries we accept and the experiences we record may become paths inherited by those who follow us. We cannot simply wait for professors, universities or technology companies to determine the answers for us. Our mistakes may become costs that future students are forced to bear.

The first generation to grow academically alongside AI may therefore carry a particular intergenerational responsibility. This responsibility does not require us to produce a perfect set of rules immediately. Nor does it require every student to become a technical expert. It begins with observing and documenting our own practices honestly.

Students must also participate actively in the formation of university policies rather than being trapped in a game of prohibition, use and concealment. Our experiences should not be treated only as risks to be managed. They are also an important source of knowledge for the development of educational policy.

Our generation is writing the first draft of what it means to learn with AI. Precisely because it is only a first draft, we have a responsibility to write it carefully, discuss it openly and allow those who come after us to revise it.

Using AI does not necessarily mean surrendering our agency. Refusing AI does not necessarily make us independent. What matters is that, as technology continues to extend our capacities, we still know why we are using it, where it is taking us and how to walk under our own power.

Disclosure of generative AI use: This essay was originally conceived and drafted by the author in Chinese. Generative AI was used to assist with its translation into English, and to suggest edits for clarity and concision. The author reviewed and revised the resulting text and takes full responsibility for the final content.