--- title: Maybe we’re wrong about AI description: A new study shows that public perceptions are wide of the mark --- [![logo (2)](https://www.hbmadvisory.com/hubfs/Assets%20HBM%20Theme/Images/logo%20(2).svg)](https://www.hbmadvisory.com/) [![logo (3)](https://www.hbmadvisory.com/hubfs/Assets%20HBM%20Theme/Images/logo%20(3).svg)](https://www.hbmadvisory.com/) - [Home](https://www.hbmadvisory.com/en-gb/) - [Advisory](https://www.hbmadvisory.com/advisory) - [121 Consulting](https://www.hbmadvisory.com/121-consulting) - [Content Review](https://www.hbmadvisory.com/contentreview) - [Testimonials](https://www.hbmadvisory.com/testimonial) - [Contact](https://www.hbmadvisory.com/contact-us) # Maybe we’re wrong about AI [Alan Hunter](https://www.hbmadvisory.com/media-hunter/author/alan-hunter) 01 May 2026 7 min read **A new study shows that public perceptions are wide of the mark** New technology provokes strong reactions, some of which will seem foolish in years to come. I didn’t buy a mobile phone when they began their journey to ubiquity in the early 1990s. I didn’t get one for a year or so after my friends because I thought they were corrosive. Not because of screen time – these were very dumb phones – but because in my estimation land lines worked perfectly well, thank you very much, and mobiles would just mean friends would show up late, or not at all, for a night out if they could just call at the last minute. I was – I am – a bit of a stickler for being on time. We all have our neuroses. My anti-phone thing seems ridiculous 35 years on and I am, of course, now as addicted as everyone else to the supercomputer in our pockets. Now some people are having a similar reaction to AI. A recent study, [The Impact of AI-generated Text on the Internet](https://ai-on-the-internet.github.io/), by researchers at Imperial College, London, and Stanford University makes this very clear. They sought to test whether public perception of what AI had done to the internet – made it less trustworthy, less diverse in ideas, less linguistically varied, more artificially upbeat (“You’re absolutely right. Great catch!”) – was matched in reality. First, they surveyed a group of American adults to find out whether their attitudes matched anecdotal public responses to the AI-mediated internet – they did – and then they assessed whether they were right or not. The conclusions were interesting. As the researchers drily remarked: “Notably, our findings diverge from public perception of AI's impact on the internet. The headline that people took from the research related to their discovery that 35% of newly published websites in May last year were either fully AI-generated (17.6%) or AI-assisted. That's up from zero less than four years ago. The more interesting finding is that our view of AI and its effects is largely erroneous. I say largely because the study confirmed that AI has made the internet more samey, in terms of the ideas and viewpoints expressed, and more artificially positive. However, by a number of other measures, our perception does not match reality. Take the big one: that AI means there is more factually incorrect information online and more hallucinations. Many journalists console themselves that this is the case – one publication is even running a blog outlining AI-based journalistic mistakes (though not one, I note, for human-originated mistakes, which will be far greater in number). Indeed, some 65% of those surveyed believed this to be the case too. Yet the study found there was no evidence that the truthfulness of the online world has declined with the advent of AI. It also dashed the idea that it has brought a more uniform style of writing – even though 80% of people thought it had. It said that there was not an increase in longer writing with less semantic density – aka dumbing down – and nor was information being posted with fewer links out to other sites as citations. In short, this research says AI is not making the internet a materially “worse” place, as many people fear it is. What should news publishers take from this? First, we must acknowledge that if 35% of the internet is bot-built and if it’s pretty trustworthy, smart and well sourced – despite what people think – then we have a really formidable competitor on our hands. Publishers need to be very clear about what they can do to differentiate themselves from this content. It’s quite a challenge, but we have things as humans that the bots don’t: contacts, presence, imagination. We need to use them. Second, we have to act fast. There’s no point in May 2026 trying to hold on to the types of stories that bots will do better. So either use bots yourself to produce that commoditised content, or accept that you will have to do something different. But you need to make that choice now, because if you put it off it’ll be too late. Third, expect public perceptions about AI to change quickly, as mine did with mobile phones. A detail in the research was that those who used AI the least among the survey sample were the most opposed to it. We can expect people’s attitudes to be more positive as its usage increases. So my guess is that there are huge opportunities for those publishers who are willing to fully embrace what AI can bring. This feels really exciting, both for companies and for individual journalists, for whom the barrier to entry to becoming publishers themselves is now much lower. But that’s for another day… ## Next blog post ## [Why it's so hard to publish fewer stories](https://www.hbmadvisory.com/media-hunter/why-its-so-hard-to-publish-fewer-stories) [Read more](https://www.hbmadvisory.com/media-hunter/why-its-so-hard-to-publish-fewer-stories) ![logo (2)](https://www.hbmadvisory.com/hubfs/Assets%20HBM%20Theme/Images/logo%20(2).svg) Hunter Brunt Media Ltd. Company number: 13546179 Registered address: Aston House, Cornwall Avenue, Church End, London N3 1LF - [Privacy Policy](https://www.hbmadvisory.com/privacy-policy) ```json { "@context" : "https://schema.org", "@type" : "BlogPosting", "author" : { "@type" : "Person", "name" : "Alan Hunter", "url" : "https://www.hbmadvisory.com/media-hunter/author/alan-hunter" }, "dateModified" : "2026-11-06T17:13:24.132Z", "datePublished" : "2026-05-01T16:00:00.000Z", "headline" : "Maybe we’re wrong about AI", "mainEntityOfPage" : { "@id" : "https://www.hbmadvisory.com/media-hunter/maybe-were-wrong-about-ai", "@type" : "WebPage" }, "publisher" : { "@type" : "Organization", "logo" : { "@type" : "ImageObject" } } } ```