Guardian columnist Zoe Williams: the AI conversation can no longer keep up with the technology itself

Guardian columnist Zoe Williams describes the strange state the public conversation about AI has fallen into: everyone talks about it, but almost no one can keep pace with how fast the technology itself is changing. As a result, even the loudest arguments sound outdated before they have finished.

Fatigue with the topic

The piece opens with an everyday scene: at an 80th birthday party, the author meets a scientist who does not immediately admit that he works in computer science. The reason is simple and almost comic: he does not want to start yet another conversation about artificial intelligence. For Williams, this instantly sets the tone for the whole column. AI is argued about so much that the topic has become tiring even for the people closest to it.

In her view, the problem is not only the quantity of conversation but its quality. Almost every position, whether euphoric or apocalyptic, lags behind reality. While some repeat that new technologies always frighten people at first and then cure cancer and move civilization forward, the technology has already changed its scale, interface and field of application. As a result, public debate increasingly sounds like a play about the future written from yesterday.

Speed versus reflection

The column's central idea is that AI is developing faster than society can make sense of it. You can demand freedom for developers, or insist on strict regulation, but both positions often arrive late. While people debate what machines should not be entrusted with, models are already getting better at that very task. So an ordinary exchange of opinions begins to feel not like control but like capitulation to the pace of change.

"We will be destroyed by a craving for endless chatter that leads nowhere."

Williams does not argue that discussion is unnecessary. On the contrary, she acknowledges that reading, learning and arguing are essential. But there is an unpleasant paradox here too. We are used to thinking of conversation as a way to work out a solution collectively, yet with AI it keeps stalling, splits into camps, and gives no sense that anyone is truly in control. Even thoughtful discussion loses to the speed of a system that reads, summarizes and changes faster than a human can.

Where the risks show

The author lists not one universal threat but a whole set of fault lines. In her account, criticism of AI covers both philosophical fears about the end of human creativity and very practical questions of employment, power and infrastructure. So the argument is not a choice between "technology will save the world" and "technology will destroy everyone" but a matter of many concrete consequences that are already starting to appear.

• pressure on service professions and routine office work
• fewer entry-level jobs for young professionals
• automatic screening of résumés and applications before any human contact
• the environmental cost of computing and data centers
• military and politically repressive uses of AI

The most unsettling part of this picture is that all these issues demand urgent attention, while the mechanisms of public response remain slow. You can discuss the interests of the billionaires who control key AI companies, or dispute the real scale of the environmental damage, as much as you like, but the sense of being behind never goes away. By the time the argument becomes substantive, its subject has already changed.

What this means

The Guardian column does not offer an action plan, but it captures the mood of 2026 precisely: society is tired of talking about AI and at the same time cannot stop. For the industry, regulators and ordinary users, it is a signal that the old pace of discussion is no longer enough. What is needed is a faster, more practical and more concrete conversation about where the technology is genuinely useful and where it must be restricted before it steps ahead again.