The risks of inviting AI into the heart of our economy, society and governance | AI (artificial intelligence)

The risks of inviting AI into the heart of our economy, society and governance | AI (artificial intelligence)

The risks of inviting AI into the heart of our economy, society and governance | AI (artificial intelligence)

https://www.theguardian.com/technology/2026/jun/03/the-risks-of-inviting-ai-into-the-heart-of-our-economy-society-and-governance

Publish Date: 2026-06-03 12:38:00

Source Domain: www.theguardian.com

Nesrine Malik is right to worry about the effect that AI may have on writing (AI is devoid of meaning and humanity. That’s why its vapid voice suits this political moment, 1 June). The examples she cites of fabricated quotations and unreliable research should concern anyone who values truth and public trust.

However, I suspect the deeper problem is not AI’s bland prose but its relationship to evidence. The writers caught out by false quotations were often not trying to deceive. They believed that they were using AI as a research aid while retaining editorial control. Yet somehow, fiction entered the factual record. The issue was not laziness but misplaced confidence in a system that can produce plausible reconstructions without distinguishing between what was observed, inferred or simply generated.

Malik is surely right that overreliance on AI may weaken habits of thought and expression. But the greater danger is that it weakens our habit of checking where ideas came from in the first place. A fluent sentence is not the same as a trustworthy one. The challenge posed by AI is not merely preserving individual voice, important though that is. It is preserving provenance – our ability to trace claims back to evidence and observation. A fabricated quote and a genuine one can now arrive wearing the same suit.
Dr Simon Nieder
Chesterfield, Derbyshire

An excellent column by Nesrine Malik. I first came across AI in 1989 while doing a feasibility study for a science park linked to Edinburgh University, which had a research institute devoted to AI. Although interested, I was sceptical of its utility. Developments since have increased my scepticism and converted it to deep concern.

As Malik observes, AI’s dependence on large language models makes the language it uses glib and persuasive, but the ideas it clothes (even when free of gross errors) are necessarily derivative and unoriginal. And as more of what it is trained on itself becomes AI-generated, this…

Source