AI, LinkedIn and the danger of becoming invisible

There’s been a lot of chatter on my LinkedIn feed of late about companies using large language models (LLMs) to generate content. Depending on your views about AI, this is either a fantastic productivity boost or the moment where Skynet becomes self-aware.

This post is neither of those things, and it’s certainly not a Luddite rant about hyphens, serial commas, or antithesis.

Because ultimately, who cares if a CEO emails a few half-formed ideas to the office admin and asks them to run it through ChatGPT to produce a “thought leadership” post for LinkedIn? AI is a tool, and like every other communications tool that came before it, it will inevitably become part of the normal workflow.

But there are risks in using it carelessly, particularly if you are trying to position yourself — or your organisation — as a credible voice in a crowded field.

The first risk is not that people will spot you’re using AI (trust me, they will).

No. The real risk is that they simply won’t care. And they won’t care because your AI-generated slop looks exactly like the AI-generated slop filling the rest of their feed. The tone is familiar. The structure is familiar. The observations are technically correct but strangely interchangeable.

So they scroll past.

At first glance that might seem harmless. After all, LinkedIn is full of content and very little of it receives sustained attention. But if your’e presenting yourself as a thought leader — someone whose analysis is worth paying attention to — congratulations, your “thinking” has now been commodified.

And commodities, by definition, are fungible. They can fit anywhere — or nowhere.

Once your ideas become interchangeable with everyone else’s, they stop carrying the signal you were hoping to send in the first place: that you have a perspective worth paying attention to. And that matters more than you might think.

At this point, many senior leaders will be thinking: “Why yes, there is a lot of generic content on LinkedIn … but my posts are different.”

And that may well be true. But the issue here is that reputation carries weight, and a CEO’s perspective on an industry issue inevitably attracts more attention than the same observation from someone more junior. But that attention is a double-edged sword.

When someone with seniority publishes something thoughtful, readers assume it reflects judgement built through years of experience and expect insight that goes beyond the obvious. When the same leader publishes something that reads like every other LinkedIn post in the feed, the contrast with hard-thought analysis becomes obvious.

The people who get invited to keynote conferences, sit on panels, or provide strategic advice are rarely the ones producing the largest volume of content. They are the ones who develop a recognisable voice and a clear point of view. They say things that feel grounded in experience rather than assembled from a collection of trite observations.

When someone is choosing between speakers, contributors, or advisers, they are effectively making a judgement about authority. So do they choose the original voice that pushes boundaries and articulates a clear position? Or do they go with one of the eleventy-thousand other “thought leaders” telling you this isn’t about productivity. It’s about focus.

You know the answer.

None of this means AI has no role in the process. Used well, it can be extremely helpful. It can help structure an argument, test the clarity of a paragraph, or pressure-test an idea. In that sense it’s no different from the spell-checkers, grammar tools and editing software that have quietly assisted writers for decades.

The difference is that AI can now generate an entire piece of writing in seconds. And while that capability is impressive, it also creates a temptation to outsource the thinking itself. And when that happens, we have a problem, because thought leadership is about demonstrating judgement, not just churning out enough words to game an algorithm or tick a “thought leadership” box on your social media “to-do” list.

And judgement is much harder to automate.

I’ve written before about the parallels between the early days of the desktop publishing revolution and the arrival of AI. When those Silicon Valley stars aligned and democratised publishing, they also opened the floodgates to an awful lot of design that should never have seen the light of day. (I was definitely guilty of some crimes in those days.)

Suddenly everyone could produce newsletters, brochures and posters with stretched fonts and clip art of questionable quality. It was the Wild West of design where the tools made publishing easier, but they didn’t automatically produce good design or sound judgement.

Over time, the novelty faded and designers who understood hierarchy, restraint and typography continued to stand out. The rest quietly disappeared. Passive-aggressive notices in Comic Sans still appear on the office kitchen noticeboard, but thankfully it’s nothing like the late 1980s and early 1990s.

Right now we’re in the equivalent phase of the AI cycle — the “just because you can, doesn’t mean you should” stage where the platforms are fast filling up with perfectly fluent, structurally sound and completely forgettable content.

Eventually, that too will pass.

The people who emerge from this melange of anodyne corporate crap will be the ones who bring something that can’t easily be replicated: a distinctive perspective and the confidence to take a position that may not be universally shared.

In other words, the same qualities that allowed good designers to survive the desktop publishing revolution.

Everyone else will fade quietly into obscurity.

Which is not really where you want to end up if you’re presenting yourself as a thought leader.