It started with a comment about music.
@JEREMY HAWKE , a courier business owner from Exeter and long-standing UKBF member, mentioned in passing that he'd been listening to something he enjoyed this week — only for his granddaughter to point out it was AI-generated. A small moment, easily dismissed. Except it wasn't dismissed. It opened a thread-within-a-thread that touched on something most of us have noticed but perhaps haven't quite put into words yet.
The internet is starting to feel different. And not in a good way.
The flood nobody voted for
When people talk about AI content, the conversation usually focuses on the tools: ChatGPT for writing, Midjourney for images, and various platforms for video. What gets less attention is the cumulative effect of all those tools being used simultaneously by millions of people to fill every available space online with output that is fast, cheap, and often indistinguishable from the real thing at a glance.
I've experienced this myself on YouTube. You find a video on a topic you're interested in, settle in, and then slowly realise that the voiceover is one of a handful of generic AI voices that now narrate a significant proportion of the platform's content. The video exists not because someone had something to say, but because AI made it cheap to produce something that would attract a click. Educational and investigative content that I used to enjoy and seek out has, in many cases, been buried under a rising tide of extended AI-generated filler.
@DontAsk, another member added a detail that will be familiar to anyone who has fallen into this trap: pictures and video footage that bear no relation to what the voiceover actually says, repeated in loops just to pad the runtime.
This isn't a minor aesthetic problem. It is a signal degradation problem. The internet's value as a source of information depends on the ratio of signal to noise. When AI makes it trivially easy to generate noise at scale, that ratio shifts — and not gradually.
LinkedIn and the unchecked share
The same dynamic is playing out on professional networks. @MikeJ , a member from Northumberland, described seeing AI-generated posts on LinkedIn that used the same abbreviation three different ways within a single post — an error that anyone reading carefully would catch immediately. The posts were being reshared enthusiastically regardless.
This is a specific kind of problem that deserves its own name. It isn't just that the content is wrong; it's that the people sharing it haven't read it. The AI generates something that looks plausible, the person posts it because it saves them time, someone else shares it because it looks credible, and the error propagates through a network of people, none of whom stopped to actually check whether it made sense.
For business owners, this matters in a concrete way. LinkedIn is still the primary professional networking platform for most UK businesses. If the quality of content on it is degrading, if the posts being recommended to you are increasingly AI-generated placeholders rather than genuine professional insight, the platform becomes less useful. Not broken, but measurably less worth the time you're putting into it.
The music test: can you tell?
The music question is worth sitting with for a moment, because it gets at something more fundamental than content quality.
Jeremy Hawke liked what he was hearing. His granddaughter had to tell him it was AI. I've had exactly the same experience — found something worth saving, went to add it to a Spotify playlist, and discovered I couldn't because it was AI-generated content on YouTube rather than a track with a proper release.
The experience of not being able to tell is not, in itself, a failure. If the music is enjoyable, the enjoyment is real. But there's something slightly disorienting about it — a version of the uncanny valley, not for appearance but for authenticity. You thought you were connecting with something made by a person. You weren't. And when you find out, it changes how you feel about it, even if you can't quite explain why.
As I explored in a previous article on AI, virtual influencers and copyright, the legal landscape around AI-generated music is still evolving rapidly. But the cultural question — what does it mean to consume content when you can no longer reliably tell whether a human made it — is one we're all navigating in real time, without a map.
The content marketing problem
For small business owners who rely on content to reach customers, this shift creates a specific strategic challenge.
If you've invested in a blog, a newsletter, a YouTube channel, or a LinkedIn presence, you are now competing for attention in a space where the volume of content is increasing exponentially while the time people have to consume it is not. AI content doesn't just dilute the pool — it trains audiences to scroll faster, trust less, and engage less deeply. Why read carefully if most of what you encounter isn't worth careful reading?
The irony is that the antidote to AI content is precisely what AI cannot reliably produce: genuine experience, specific knowledge, earned credibility, and a recognisable human voice. @UKSBD , a long-standing UKBF moderator, touched on this when describing how he uses an AI content plugin to generate blog post drafts — but edits everything before publishing, adds images, adds internal links, and rewrites wherever the output doesn't reflect what he actually knows. The AI is doing the rough cut; the value is coming from the human finishing it.
That approach — AI as a starting point, not an endpoint — is probably the most honest account of where the technology is genuinely useful for content creation. And I'll be transparent: it's exactly what I've done with this article. I worked with Claude to produce the first draft, then carefully read it through, tweaked the tone and phrasing, added links to member profiles, incorporated references to previous articles, and shaped it to reflect my own perspective and voice. The AI did the initial heavy lifting; the judgment, the context, and the accountability are mine. If that sounds like a reasonable way to work, it is. But it requires discipline and time. The temptation to skip the editing step and just publish the raw output is exactly how noise gets made.
Does volume destroy value?
@fisicx, the thread's originator, raised a broader structural concern underlying all of this: the economics of AI content generation are deeply unstable.
The current wave of AI-generated content is being produced at a price point that, in many cases, is close to zero. The tools are cheap or free, the marginal cost of another piece of content is negligible, and the barrier to publishing is essentially non-existent. This means the volume of AI content will continue to grow until something stops it — either the platforms develop better filtering, audiences learn to ignore it more systematically, or the economics change.
His specific concern about the environmental and financial sustainability of AI infrastructure is a longer conversation, but the content problem has a more immediate dimension: when everything can be made, what signals quality? When anyone can publish anything about anything at no cost, how do you find what's actually worth reading?
The answer, for now, appears to be: reputation, consistency, and the kind of specific credibility that only comes from actually knowing what you're talking about. A business owner who has been writing honestly about their industry for five years has something that cannot be generated. A community like UKBF, where contributions are made by real people accountable to each other over years or decades, has something that cannot be faked.
What this means practically
If you're a business owner thinking about your own content and online presence, a few things are worth bearing in mind.
Be sceptical of what you share. The LinkedIn problem isn't going away, and being the person who reshares AI-generated nonsense is a reputational risk in a way it wasn't two years ago. Read before you share. Check before you endorse.
Lean into specificity. AI content tends towards the general because it's trained on the general. Content that draws on specific experience — a particular client situation, a specific market insight, a real mistake and what you learned from it — is both harder to fake and more valuable to read.
Don't compete on volume. If you're producing content to be heard, the answer to AI volume isn't more volume. Its distinctiveness. One piece of content that is genuinely useful and clearly from a real person with real expertise will outperform ten AI-generated pieces in terms of the outcomes that actually matter: trust, enquiries, and reputation.
And perhaps most importantly: pay attention to the signal-to-noise ratio in the sources you rely on. The information environment you operate in affects the quality of your decisions. If the feeds you follow are filling with AI slop, it's worth being deliberate about curating them back towards sources you've learned to trust.
The question isn't whether AI content exists. It's what you do about it.
The internet has survived other floods of low-quality content. It survived the SEO content farms of the early 2010s, the clickbait era, and the social media engagement-bait cycle. It tends to adapt, if slowly and imperfectly.
But the speed and scale of AI content generation is qualitatively different from those earlier waves, and the adaptation this time may require more active effort from those of us who depend on the internet to run our businesses.
The thread on UKBF that sparked this article is, in its own small way, an example of what the alternative looks like: real people, named and accountable, sharing genuine observations from lived experience, debating honestly and without a script. That kind of exchange doesn't scale as well as AI-generated content does. But it's worth considerably more.
This article draws on the "AI does have its uses" discussion thread on UK Business Forums, with contributions from members @JEREMY HAWKE, @DontAsk, @MikeJ, @fisicx, and @UKSBD.
Related reading: The AI Divide: What UKBF Members Really Think in 2026 | When the Algorithm Gets It Wrong
@JEREMY HAWKE , a courier business owner from Exeter and long-standing UKBF member, mentioned in passing that he'd been listening to something he enjoyed this week — only for his granddaughter to point out it was AI-generated. A small moment, easily dismissed. Except it wasn't dismissed. It opened a thread-within-a-thread that touched on something most of us have noticed but perhaps haven't quite put into words yet.
The internet is starting to feel different. And not in a good way.
The flood nobody voted for
When people talk about AI content, the conversation usually focuses on the tools: ChatGPT for writing, Midjourney for images, and various platforms for video. What gets less attention is the cumulative effect of all those tools being used simultaneously by millions of people to fill every available space online with output that is fast, cheap, and often indistinguishable from the real thing at a glance.
I've experienced this myself on YouTube. You find a video on a topic you're interested in, settle in, and then slowly realise that the voiceover is one of a handful of generic AI voices that now narrate a significant proportion of the platform's content. The video exists not because someone had something to say, but because AI made it cheap to produce something that would attract a click. Educational and investigative content that I used to enjoy and seek out has, in many cases, been buried under a rising tide of extended AI-generated filler.
@DontAsk, another member added a detail that will be familiar to anyone who has fallen into this trap: pictures and video footage that bear no relation to what the voiceover actually says, repeated in loops just to pad the runtime.
This isn't a minor aesthetic problem. It is a signal degradation problem. The internet's value as a source of information depends on the ratio of signal to noise. When AI makes it trivially easy to generate noise at scale, that ratio shifts — and not gradually.
LinkedIn and the unchecked share
The same dynamic is playing out on professional networks. @MikeJ , a member from Northumberland, described seeing AI-generated posts on LinkedIn that used the same abbreviation three different ways within a single post — an error that anyone reading carefully would catch immediately. The posts were being reshared enthusiastically regardless.
This is a specific kind of problem that deserves its own name. It isn't just that the content is wrong; it's that the people sharing it haven't read it. The AI generates something that looks plausible, the person posts it because it saves them time, someone else shares it because it looks credible, and the error propagates through a network of people, none of whom stopped to actually check whether it made sense.
For business owners, this matters in a concrete way. LinkedIn is still the primary professional networking platform for most UK businesses. If the quality of content on it is degrading, if the posts being recommended to you are increasingly AI-generated placeholders rather than genuine professional insight, the platform becomes less useful. Not broken, but measurably less worth the time you're putting into it.
The music test: can you tell?
The music question is worth sitting with for a moment, because it gets at something more fundamental than content quality.
Jeremy Hawke liked what he was hearing. His granddaughter had to tell him it was AI. I've had exactly the same experience — found something worth saving, went to add it to a Spotify playlist, and discovered I couldn't because it was AI-generated content on YouTube rather than a track with a proper release.
The experience of not being able to tell is not, in itself, a failure. If the music is enjoyable, the enjoyment is real. But there's something slightly disorienting about it — a version of the uncanny valley, not for appearance but for authenticity. You thought you were connecting with something made by a person. You weren't. And when you find out, it changes how you feel about it, even if you can't quite explain why.
As I explored in a previous article on AI, virtual influencers and copyright, the legal landscape around AI-generated music is still evolving rapidly. But the cultural question — what does it mean to consume content when you can no longer reliably tell whether a human made it — is one we're all navigating in real time, without a map.
The content marketing problem
For small business owners who rely on content to reach customers, this shift creates a specific strategic challenge.
If you've invested in a blog, a newsletter, a YouTube channel, or a LinkedIn presence, you are now competing for attention in a space where the volume of content is increasing exponentially while the time people have to consume it is not. AI content doesn't just dilute the pool — it trains audiences to scroll faster, trust less, and engage less deeply. Why read carefully if most of what you encounter isn't worth careful reading?
The irony is that the antidote to AI content is precisely what AI cannot reliably produce: genuine experience, specific knowledge, earned credibility, and a recognisable human voice. @UKSBD , a long-standing UKBF moderator, touched on this when describing how he uses an AI content plugin to generate blog post drafts — but edits everything before publishing, adds images, adds internal links, and rewrites wherever the output doesn't reflect what he actually knows. The AI is doing the rough cut; the value is coming from the human finishing it.
That approach — AI as a starting point, not an endpoint — is probably the most honest account of where the technology is genuinely useful for content creation. And I'll be transparent: it's exactly what I've done with this article. I worked with Claude to produce the first draft, then carefully read it through, tweaked the tone and phrasing, added links to member profiles, incorporated references to previous articles, and shaped it to reflect my own perspective and voice. The AI did the initial heavy lifting; the judgment, the context, and the accountability are mine. If that sounds like a reasonable way to work, it is. But it requires discipline and time. The temptation to skip the editing step and just publish the raw output is exactly how noise gets made.
Does volume destroy value?
@fisicx, the thread's originator, raised a broader structural concern underlying all of this: the economics of AI content generation are deeply unstable.
The current wave of AI-generated content is being produced at a price point that, in many cases, is close to zero. The tools are cheap or free, the marginal cost of another piece of content is negligible, and the barrier to publishing is essentially non-existent. This means the volume of AI content will continue to grow until something stops it — either the platforms develop better filtering, audiences learn to ignore it more systematically, or the economics change.
His specific concern about the environmental and financial sustainability of AI infrastructure is a longer conversation, but the content problem has a more immediate dimension: when everything can be made, what signals quality? When anyone can publish anything about anything at no cost, how do you find what's actually worth reading?
The answer, for now, appears to be: reputation, consistency, and the kind of specific credibility that only comes from actually knowing what you're talking about. A business owner who has been writing honestly about their industry for five years has something that cannot be generated. A community like UKBF, where contributions are made by real people accountable to each other over years or decades, has something that cannot be faked.
What this means practically
If you're a business owner thinking about your own content and online presence, a few things are worth bearing in mind.
Be sceptical of what you share. The LinkedIn problem isn't going away, and being the person who reshares AI-generated nonsense is a reputational risk in a way it wasn't two years ago. Read before you share. Check before you endorse.
Lean into specificity. AI content tends towards the general because it's trained on the general. Content that draws on specific experience — a particular client situation, a specific market insight, a real mistake and what you learned from it — is both harder to fake and more valuable to read.
Don't compete on volume. If you're producing content to be heard, the answer to AI volume isn't more volume. Its distinctiveness. One piece of content that is genuinely useful and clearly from a real person with real expertise will outperform ten AI-generated pieces in terms of the outcomes that actually matter: trust, enquiries, and reputation.
And perhaps most importantly: pay attention to the signal-to-noise ratio in the sources you rely on. The information environment you operate in affects the quality of your decisions. If the feeds you follow are filling with AI slop, it's worth being deliberate about curating them back towards sources you've learned to trust.
The question isn't whether AI content exists. It's what you do about it.
The internet has survived other floods of low-quality content. It survived the SEO content farms of the early 2010s, the clickbait era, and the social media engagement-bait cycle. It tends to adapt, if slowly and imperfectly.
But the speed and scale of AI content generation is qualitatively different from those earlier waves, and the adaptation this time may require more active effort from those of us who depend on the internet to run our businesses.
The thread on UKBF that sparked this article is, in its own small way, an example of what the alternative looks like: real people, named and accountable, sharing genuine observations from lived experience, debating honestly and without a script. That kind of exchange doesn't scale as well as AI-generated content does. But it's worth considerably more.
This article draws on the "AI does have its uses" discussion thread on UK Business Forums, with contributions from members @JEREMY HAWKE, @DontAsk, @MikeJ, @fisicx, and @UKSBD.
Related reading: The AI Divide: What UKBF Members Really Think in 2026 | When the Algorithm Gets It Wrong
