Between Fear and Promise: Why AI Needs a More Inclusive Imagination

Photos by Massimo Pistore / University of Padua

When Payal Arora takes the stage, she tends to unsettle her audiences, not with alarm but with a kind of disarming — not naïve — optimism. A digital anthropologist, award-winning author, and professor of Inclusive AI Cultures at Utrecht University, Arora has spent two decades studying how billions of people across the Global South — in India, Brazil, China, Iran — relate to digital technology. What she found contradicts much of the received wisdom circulating in European and American policy circles: for the majority of the world's population, the internet and artificial intelligence are not primarily threats to be managed, but tools of emancipation, access, and social participation.

Arora was one of the keynote speakers at Rethinking the University in the Age of Artificial Intelligence, a two-day international conference held at Palazzo del Bo, the historic seat of the University of Padua, organised within the ALMA network and dedicated to the future of academic teaching in the age of generative AI. Her contribution, titled Teaching AI to See the World Differently, drew on her latest book, From Pessimism to Promise (MIT Press), and on the work of the Inclusive AI Lab, the international research network she co-founded, which brings together researchers, civic organisations, and business voices from across the Global South to co-design AI systems that reflect the world's true diversity.

Against the backdrop of geopolitical tensions and technological acceleration, Arora warned in her speech of a growing inward turn in Europe, a “fortress” mentality that risks narrowing both education and imagination. Too often, she argued, public discourse oscillates between technological arrogance and social pessimism — overestimating what machines can do while underestimating the role of people and institutions. “We keep repeating the same narrative: that new technologies will make us stupid. Years ago it was Google; today it’s AI. Each time, the technology changes, but the fear remains remarkably similar, often without strong empirical grounding. Meanwhile, much of the rest of the world sees AI as an opportunity to democratise, to connect, to open up knowledge. And yet here we see emerging almost a ‘puritan’ stance — refusing to use AI as a badge of integrity.”

We spoke with her on the margins of the conference, in a conversation that shifted from classroom to family dinner table, from Tehran to Silicon Valley, and from the anxieties of middle-aged smartphone users to the quietly subversive habits of Generation Z.

Q. You argue that fears of AI replacing human activities, such as teaching, are in some ways misplaced. Why?

Payal Arora: Every technology is intrinsically assistive: AI won’t replace human teaching, just as it won’t replace parenting or governing, or any domain that is embedded with care. If we start framing the question in terms of replacement, we are essentially abdicating responsibility — blaming technology for something that is fundamentally human. The real question is different, and more difficult: what is the role of our social institutions, and how can they use these tools to improve well-being — for society and for the planet? That’s a far more complex issue, and not one that easily makes headlines.

Q. Yet many parents and teachers are concerned about the impact of technology on younger generations.

Arora: I understand that concern, but there is also a tendency to underestimate what young people are actually doing with technology. We are constantly exposed to moral panics — stories about harmful behaviours, extreme cases — which create anxiety and make parents feel they must do everything possible to protect their children.

The data, however, tell a different story. Empirical research shows that it is often older generations — people between 45 and 65 — who are most attached to their phones. Younger people, in many contexts, find it “uncool” to be constantly online; they even reprimand adults for it. In a sense, we are projecting our own behaviours onto them, but they are a different generation, and many are consciously trying not to replicate our habits.

Q. Is artificial intelligence, then, more a force for inclusion or a source of bias and discrimination?

Arora: That’s like asking whether a university or a family is good or bad: it depends! A family can be nurturing or abusive; a university can expand knowledge or constrain it. Why do we hold technology to a completely different standard?

Technology reflects the full spectrum of human capability — from the darkest uses, like deepfake pornography or child exploitation, to deeply enabling ones, such as giving people with disabilities new ways to participate in everyday life. And we should remember that, at some point, all of us will experience forms of limitation — physical or mental — where technology can become a lifesaver and a lifeline to social participation.

So AI is not inherently one thing or another. It is a social institution, and we should treat it as such: building systems of accountability and care, while also allowing it to develop its potential. Over time, like past technologies, it may move from being seen as radical to becoming something more ordinary — integrated into everyday life.

Photos by Massimo Pistore / University of Padua

Q. You’ve highlighted how perspectives differ globally. In parts of Asia and the Global South, technology is often seen more optimistically. But there are also clear risks — control, concentration of power. How should we think about that?

Arora: We should take those risks seriously, but also look at how people respond to them. Take Iran, for example: when governments shut down the internet — as has happened repeatedly — it becomes clear how powerful technology is. Internet shutdowns are now one of the most common tools of control, precisely because access to technology enables knowledge, connection, and collective action.

And yet, young people in those contexts are not asking to be cut off. They want access — they want in. That should tell us something. At the same time, wanting access does not mean “anything goes”: It also implies a demand for accountability. So again, it’s not a simple binary: it’s about recognizing both the risks and the transformative potential.

Q. In Western democracies, too, there are concerns about control over information, media, and even political processes. Is this a growing threat?

Arora: Of course there are new challenges and threats. But I think it helps to see our relationship with technology as just that — a relationship. Like a marriage: there is commitment, but also work. Technology has a role in our lives, but we need to define what that role is and continually renegotiate it.

We already rely on these tools in many ways: to organize our lives, to motivate ourselves, even to reflect on our own thinking. Sometimes technology can help us see beyond our own “filter bubbles,” to recognize the limits of our perspectives. It can support personal growth — but only if we are intentional about how we use it.

Q. In your talk, you said that we need to shift from “training AI” to “teaching AI”, going beyond data because “garbage in, garbage out”. How can we make generative AI more open and inclusive?

Arora: The first step is to actively look for bias. When you use these tools, pay attention to what they produce. It’s similar to a Google image search: type in “family,” and you’ll often see a very narrow representation — two parents, two children, smiling. But that’s only one version of reality, because there's a certain norm of what a family is supposed to look like. At the same time, these systems can also be used as extraordinarily powerful tools: a kind of collective, living archive of the plurality of the world.

Generative AI works in the same way: if it reflects only a small slice of that world, we should ask why. Where do these representations come from? What data is being used? These questions matter because the outputs we see feed back into the datasets that train future systems.

In that sense, users themselves play a role. By interrogating what we see — by questioning norms and representations — we can push these systems toward a broader, more inclusive understanding of the world.