Just because we can, doesn't mean we should

Jan Broersen paints an optimistic picture of AI as a tool in the fight against climate change. According to him, large language models could better inform people and even positively influence political decision-making. That vision sounds appealing, but it is based on problematic assumptions and ignores serious ethical objections.

Broersen's arguments
Broersen argues that AI can be more reliable than individual humans because large language models are trained on enormous amounts of text, thereby smoothing out extreme viewpoints. He also argues that AI is difficult to manipulate, as evidenced, in his view, by the example of the chatbot Grok, which initially adhered to the scientific consensus. Finally, he hopes that chatbots can steer citizens and politicians towards more ambitious climate policies.

An illusion of objectivity
However, this reasoning is flawed. AI is not objective, as it reproduces patterns from the data it's been trained on. If that data is biased, the bias is reinforced, not corrected. "The truth" is not embedded in statistics. Moreover, the Grok example undermines Broersen's point: ultimately, the system has indeed been modified and is now more receptive to climate denial. This demonstrates exactly how AI is influenced by powerful players, namely the tech companies that Broersen himself is concerned about.

The assumption that better information automatically leads to better beliefs is also naïve. Climate denial is not a knowledge deficit, but rather a psychological and political problem. People deny climate change because acknowledging it conflicts with their identity, interests or worldview. A chatbot does not change that.

Ethical problems that Broersen fails to mention
Broersen also pays too little attention to the ethical costs of AI. The vast amounts of energy and water that data centres consume are not a side issue, but a structural problem. It is paradoxical to present a technology that itself contributes to climate change as a solution.

Added to this are large-scale copyright infringements. Language models are trained on millions of works without permission or compensation, while they can subsequently displace the creators. This is fundamentally problematic not only from a legal point of view but also from an ethical point of view. Moreover, academic integrity is under pressure when students and researchers outsource their thinking to systems that generate superficial answers.

The concentration of power in the hands of a handful of tech companies poses an additional risk. By presenting AI as a neutral source of knowledge, we are giving these companies even more influence over the public debate. At the same time, epistemic authority is shifting toward black-box systems whose functioning is largely opaque, undermining critical thinking and democratic debate. On top of that, AI has privacy issues: every interaction with AI generates data that is commercialised.

Conclusion: doing good is not enough
In his conclusion, Broersen argues for optimism, but optimism should not become techno-utopia. The end does not justify the means. Even a noble goal such as climate justice does not justify technology that itself damages the climate, democracy, creativity and human autonomy.

That does not mean that technology cannot play a role by definition. I share that hope. But current AI systems are built on exploitation, high ecological costs and concentration of power. Moreover, Broersen overestimates what AI can achieve. More information does not automatically lead to different behaviour. The climate crisis calls for political choices, social change and collective responsibility, not for outsourcing our thinking to machines. The core issue is simple: just because something is technically possible does not mean it is desirable. 

There are better, more democratic and less harmful ways to inform people about climate change, such as investing in education, journalism and science communication. Not every solution deserves to be tried, especially when the cure is more dangerous than the disease.