“Bots meet voters” — The Economist’s 4 strategies to win public buy-in for AI

In Bots meet voters, The Economist likens AI’s potential to change the world to that of electricity and the steam engine. It argues this new world-changing technology is facing poor public buy-in, and that can be fixed through four strategies: spreading AI’s benefits more widely, regulating its sharpest dangers, measuring its resource consumption accurately, and using AI to make states more efficient.

These are sensible suggestions. But they are built on an assumption: that the public mistrust of AI is mainly a problem of perception, and that can be overcome with some philanthropy and better numbers. That assumption may have held for technologies that changed what people could do with their hands and time. It is less convincing for technology that competes with the human mind itself — at least not yet convincing. If AI reshapes not only labour but also the kinds of intelligence societies value, then the challenge is not merely one of public trust but also of education and social preparedness. Let us look at the paper’s strategies a little more closely.

Spreading AI’s benefits as widely as possible. The paper offers the example of data centres facing local resistance and it suggests tech companies should do some local philanthropy so that communities feel they have economic stake in AI’s progress. That’s a reasonable fix because data centres are really not much different from power stations. But the resistance that AI faces is not local in a geographic sense only. The fear that AI is gobbling up jobs and that companies are laying off people in anticipation of what AI can do constitutes a greater angst than something some corporate social responsibility can fix. Communities may welcome new investment while still fearing the long-term erosion of meaningful work.

Regulate hard against the sharpest threats of AI. This is perhaps the most urgent call. Preventing AI-aided terrorism, hacking, and other forms of malicious use is worth doing regardless of its effect on public opinion. But these risks are only a part of the bundle of risks; they are identifiable, and legislation can target them. But the way AI is re-shaping human cognitive labour subtly is a greater long-term risk that can’t be targeted by laws. This is where AI differs sharply from electricity or the steam engine. Those technologies transformed how people did things; AI increasingly performs tasks that were once thought to require human judgment. In many domains it already rivals, or encourages employers to believe it rivals, average human workers. The question is not only how to regulate malicious use, but how societies should respond when the value of human cognition itself begins to change.

Measure everything. This is persuasive. The point about resource usage is well taken. Data centers, for example, consume less water, in average, than many industries that attract little protests do. Better statistics would genuinely defuse a confected controversy.

But this strategy works precisely because resource consumption is measurable. Job displacement and gradual narrowing of opportunities for human cognitive work are not. So it only solves a problem on the surface.

Use AI to make the state better, so that people trust it because it improves their own healthcare and education. True, people are less likely to oppose a technology behind their grandmother’s cancer treatment or their child’s education.

Yet even here, the question is not whether AI makes services more efficient. It is also what kind of citizens those services are preparing people to become. If AI increasingly performs cognitive work, then education cannot merely become another sector made more efficient by AI. It must also prepare people to live and flourish in a society transformed by it.

All these strategies are useful in their own right, but they do not fully address the AI’s deeper challenge. What’s needed requires going beyond winning public buy-in. It requires preparing people for what AI actually changes. The strategies do not touch these areas.

First, structurally, education systems need curricula and modes of delivery built around adaptability to labour market being reshaped by AI. The current system does not prepare people sufficiently either to protect their agency as humans or to pursue careers that can’t easily be delegated to AI bots.

Second, and more fundamentally, education needs to produce citizens who are more socially, culturally, and intellectually aware. The risks that AI poses are, as many of its own pioneers admit, uncertain. Because the AI systems remain only partially understood, regulation will always lag behind. Where the dangers cannot be precisely measured or regulated in advance, awareness of one’s own rights, responsibilities, and position within society becomes one of the few safeguards available against technologies whose creators themselves cannot fully explain and audit how they work.

The Economist’s four strategies could make AI more popular and tech companies more profitable. But that does not make them safe. States cannot simply integrate AI into public services, pass a few laws, and assume the deeper problem has been solved. They need to confront the tectonic shift underneath. AI does not just give people more working hours by providing light or make commute easier, as electricity and steam engine did; it forces people to evaluate the place of human intelligence and its relevance. So, AI is not just a nice corporate product that’s struggling to reach to the customers — as it stands, it’s a product that consumers fear is poised to balloon out in their living rooms and push them out on the street. That’s the fear that needs to be addressed honestly.