How AI will be harnessed by the untalented to remodel human excellence into a shapeless goo.

In an earlier article titled ‘Rethinking Work: The Personal and Professional Shift with AI’, I explored the potential of large language models like ChatGPT (or more specifically, proprietary tools built on LLM technology) to produce technical-to-lay materials more effectively than the average human medical writer and to translate faster and with greater accuracy than the average human translator.

This is because of the very perceptible variation in the quality of translator and medical writer. I lump both professions together because the point I am making is applicable to both and both are parallel fields, facing the same pressures. In fact, in many cases the translator and the medical writer may be one and the same person.

So, let us revisit the pyramid:

Both translators and medical writers broadly form a pyramid made up of three bands of ability. As you go down the pyramid, we find declining levels of talent and ability. The elite apex is formed by talented and rare people, always in demand. Then, there is a reasonably competent middle tier: adequate, reliable, just-so, but often struggle with technical-to-lay writing. Then there is the bottom tier. These are the opportunists, the blaggers, the CV-padders. You find here individuals that should not be in the profession at all. They are there because they couldn’t make it elsewhere. They fake, they bluff, they imitate. They talk a good game, but under-deliver.

In an earlier article, I made the case that AI might be the great leveller, the great clarifier. A kind of digital Darwinism, I wrote. It would be a pressure hose that would wash away the bottom third and part of the middle band. It would force a rethinking of competence, to reorient towards the elite. The cream would rise to the top: the truly talented would thrive, freed from the noise and mess and the zombie swarm below them.

The pyramid would then look like this:

I shared this vision with a friend, Dr Neil Jenkings, a sociologist from the University of Newcastle-upon-Tyne, UK. He would not describe himself as an expert in AI by any measure, but he has made a few important observations that seem to cut through the noise and the hysteria. For example, he was the first person I know - and I read a lot on the subject – who first pointed out years ago that designers of studies will eventually grow more comfortable with the idea of using AI-powered simulated patients in Patient Voice Research – in place of humans. So, he looked at this ‘Before’ and ‘After’ and he stated that there could be a different reality, that the opposite could be happening: AI could be harnessed by the bottom tier to flatten expertise.

When the Bottom Tier Gets Smart

Neil pointed out a simple truth: opportunists are always the first to adapt. They move quickly. They do not care about craft or even ethics. They just want output. And now they have a tool, AI, that gives them all the polish, structure, grammar, voice and tone they were never able to master themselves.

The people who were barely scraping by as writers and translators now sound pretty good. Their output is not incredible but it is not bad either. It sounds competent, fluid and organised. It is “quite good”.

AI has handed the bottom tier a mask, which they are wearing very well.

The Flip of the Pyramid

We redraw the pyramid to reflect what might now be happening:

You still have the elite and the so-so, but their expertise is faded. They are indistinguishable in the market. Everyone else is using AI-assisted tools that produce fast and good-enough content. But it is flat and indistinct.

The elite band still exists. Of course it does. There are still truly brilliant minds out there. But the problem is that they are being mistaken for everyone else. When we are dealing with a lay summary or an English-to-Khmer translation, who can really tell the difference between “great” and “perfectly adequate”?

We have entered an era where surface polish is easy to fake, and depth is too expensive to care about.

The Death of Distinction

This is the real danger of AI, not that it replaces us with machines, but that it makes the mediocre look masterful. And when everything looks masterful, mastery itself loses value.

Why pay a brilliant writer who challenges your assumptions, who reshapes your story, who brings you discomfort and friction in just the right way, when you can get 80% of the result for 5% of the cost?

Why hire the elite translator who agonises over cultural nuance and rhythm, when the AI-plus-post-editor (usually a disgruntled translator whose role has evaporated) can deliver something that reads perfectly okay?

The answer is that you don’t. Not unless you still care about excellence. And fewer and fewer are able to discern what excellence is, when everything is a uniformity of ‘just-so’.

And What of the New Blood?

There’s another layer to this. The elites retire and die out. There is nobody to replace them. No torch is handed over. No shoulders of giants to stand on anymore.

AI does not care about the entry level. There is nobody to teach apprenticeships to. There is nobody to mentor to them on the way to becoming experts.

In the olde world, you started at the bottom, but you then learned by doing. You sat next to someone who was better than you. You edited their work. You watched how they thought. You absorbed their tone, judgement, discernments and behaviours. Over many years.

But if the entry level does not need to learn, because the machine does all the heavy lifting, what pipeline is left for future experts? Maybe they will not emerge. Maybe they will never develop. Why would they, when there is no struggle, no failure and no refinement?  The bright 22-year-old never becomes the 52-year-old master, a global expert.

The pyramid does not just flatten, it collapses from within. It melts into shapeless goo, like this:

The Dangers of “Quite Good”

“Quite good” sounds harmless. It sounds like a win-win situation. After all, isn’t it good that more people can produce polished content? Is this a great levelling?

The answer to these two questions is, of course, “yes”. But the issue is that when everything is “quite good”, nothing is amazing. Over time, we forget what amazing even looks like. We lose our taste for it. Our standards dull and our expectations shrink. Look at how pervasive the ChatGPT written style is on LinkedIn posts. This is just a small example of the shapeless goo, the blob.

The irony is that AI makes the world sound better, while output is worse.

Who suffers in this? The people who used to give a damn do. These are the real professionals. These are the people who honed language like a blade. These are the ones that do not rely on prompts, but on intuition, memory, instinct and sheer ability. None of these people are celebrated. They are being priced out, ghosted and erased. This is what replacement looks like. There is no meeting where you are explicitly told that you are being laid off. It is just a slow slide into obsolescence and then erasure.

The First and Only Coat of Paint?

The current wisdom that reassures us is that the human still has a role in knowledge production. We need two coats of paint: one by the machine, the other by the human, who adds the finesse.

Yes, but there is also an uncomfortable truth. If the first coat of paint produces “good enough” output, then people will stop caring about the need for the second coat.

This is because the second coat costs time, money and expertise. They are in a hurry, after all.  The new decision-makers do not want to learn how to be brilliant. They want templates, macros and shortcuts. And the people buying the work do not want brilliance either. They just want output. And they want it now. And, so, “quite good” is really… quite fine.

So What Survives?

Returning to the three-tier pyramid of now, we are at the beginning of the revolution where the bottom can mimic the middle and the middle can crowd out the top. So, what remains?

·       Maybe niche brilliance survives. This would be work so technical, so legally or ethically sensitive that AI cannot be trusted with it yet.

·       Maybe this is a phase, where the ‘good enough’ mentality will be dumped in favour of authenticity, where we value, not only creativity but the mistakes that humans do. To err is to be human would take on new meaning and be venerated.

·       Maybe nothing survives. Maybe the pyramid flattens into a platform, all clean and slick, all polish, but no structure underneath.

We used to talk about digital Darwinism as the culling of the slow, the stupid and the obsolete. But what if evolution has turned? What if we are now selecting against the best? What if it is the competent, the original and the masterful who are being edited out? This wouldn’t be out of malice but because in the age of the shapeless goo, excellence becomes irrelevant.

Thank you for reading,

Mark Gibson

Leeds, United Kingdom, July 2025