Professor warns against a naive view of AI

Johan Hansson, Professor of Applied Physics at Luleå University of Technology, has recently had an opinion piece on artificial intelligence published in the internationally renowned journal Physics World. In the article, he highlights what he sees as serious risks linked to the rapid and uncritical development of the technology.

Physics World is published by the Institute of Physics and is read by researchers and decision-makers around the world. For Johan Hansson, the publication is a way to reach a broad audience with his reflections on the consequences of AI.

“It gives me the opportunity to highlight what I perceive as a naive attitude toward AI and to discuss the risks in a broader international context.”

In the opinion piece, he argues that many underestimate the downsides of the technology. He believes there is a widespread overconfidence in the idea that development can be fully controlled.

“The biggest misconception is that we can fully control the development and that AI is just smart software. At its core, this is about physics. Everything requires energy and resources, and AI is extremely demanding in that respect.”

Energy concerns at the center

A central part of his argument concerns energy consumption and resource use. According to Johan Hansson, too little attention is paid to the vast amounts of electricity and cooling capacity required to run advanced AI systems.

He describes AI as a highly energy-intensive technology and argues that its climate impact is not taken seriously enough in the public debate. From his perspective, all technological development must be weighed against the planet’s limited resources.

He also points to risks associated with universities and other public institutions using commercial AI tools without reflecting on dependencies and long-term consequences. Instead, he believes academia should discuss how to build competence and maintain control within its own operations.

At the same time, he emphasizes that AI can be justified in certain contexts, such as advanced research where computational power is crucial. The problem arises, in his view, when the technology is used broadly without properly analyzing the consequences.

“We need a more thoughtful and responsible use of AI. Otherwise, we risk creating problems whose consequences we cannot fully foresee,” says Johan Hansson.

• Applied physics
• Read the article here (print) pdf, 227.7 kB.
  Physics World
• Read the article here (online)