How to put people first in cyberspace

Better Tech: Putting People First in Cyberspace Francine Berman MIT Press (2026)

For decades, computer scientist Francine Berman has helped to shape the digital world. In Better Tech, the University of Massachusetts Amherst scholar makes the case that technology should serve the public interest. It is a timely manifesto for how to put people’s welfare at the centre of discussions about technological progress.

You argue that digital technologies are essential infrastructure, yet often lack safeguards. Could you elaborate?

When we think about critical infrastructure such as roads, the electrical grid and the financial system, we expect safeguards to protect the public. Those protections are a long way from being perfect. Bridges collapse, power grids fail and financial crises happen. But we still invest heavily in oversight, safety measures and regulation because these systems underpin everyday life.

Digital technologies now have a similarly central role, yet we often treat them differently. Instead of assuming they should be trustworthy and well governed, we spend a great deal of time protecting ourselves from them and questioning whether we can trust them. If we began treating digital technologies as critical infrastructure, we would be more likely to build them with safety, reliability and public benefit in mind from the outset. That would mean stronger standards and clearer rules about how these technologies are developed and used. The goal is not to eliminate risk, but to take these essential technologies seriously enough to manage that risk well.

Do search engines and social-media platforms count as critical infrastructure?

I think search engines crossed that threshold long ago. Once most information moved online, such searches became the main way that people found knowledge and made decisions. In that sense, search engines are infrastructure. We need food to sustain our bodies; and we need reliable information to sustain an informed society. The problem is that today’s information systems are governed mostly by business incentives. Search engines use algorithms that connect users with advertisers, often prioritizing content on the basis of commercial interests rather than public benefit. That does not mean that they are inherently harmful, but it does mean that their influence is too important to operate without greater transparency and accountability.

The same is true of social media. Many of the most important public announcements, political debates and cultural conversations now happen there. What was once communicated through press releases or by official institutions is increasingly communicated through social platforms. They have become essential channels for public life. Yet these platforms often lack the safeguards that we would expect from infrastructure of that importance. Users face cyberbullying and predatory behaviour. Recommendation systems can pull people into endless scrolling and narrow information environments. None of this is inevitable. These platforms were designed to maximize growth and engagement, and they have been remarkably successful at doing that. But design choices are choices. We could build systems that are both profitable and protective.

Are social-media firms taking accountability for alleged harm to users, or avoiding the issue?

Perhaps a bit of both. For example, the company Meta has come to a financial settlement with some US states that claimed its products harmed child safety. But it’s worth remembering that the firm faces various legal and regulatory challenges, so the significance of any one settlement depends on how it fits into that broader landscape. One positive development, however, is that people are increasingly demanding accountability from technology companies. What interests me most about these court cases is not the financial penalties, but the product changes Meta might be required to make. The meaningful question is whether the court settlements lead to a product redesign that reduces addictive features or gives users more control.

The changes that have been publicly announced do not go far enough. Although these are small steps, I think the fact that we are starting to hold these companies accountable means that they’re more likely to create better products in the future, which is really what we want.

Do you think, like social-media firms, artificial-intelligence companies are now designing systems that can be addictive?

I think it’s important to distinguish between AI and the products that use it. AI is a broad field. In their 2021 textbook Artificial Intelligence, computer scientists Stuart Russell and Peter Norvig describe AI along two dimensions: whether a system is designed to think or to act, and whether it aims to behave like a human or like a rational agent. That creates four broad categories of AI research. These give rise to distinct research questions, and they continue to be active areas of AI development. Unfortunately, in popular discourse, AI has become a broad umbrella term, making it harder to distinguish between applications that are genuinely useful and those that are not.

What AI risks do you think we’re paying too much and, conversely, too little attention to?

Will AI ruin the social sciences — or revolutionize them?

There are genuine risks when people start treating AI systems as if they have human qualities. We’ve seen troubling cases, especially involving young people, in which individuals have turned to chatbots for companionship and received poor or even harmful advice. In some instances, the consequences have been tragic. Those risks deserve attention.