Losing your edge to AI

What happens when AI does the work that taught us how to think

TL;DR

• AI is absorbing the entry level tasks that taught people how to evaluate quality.
• Outputs go unchecked because the people commissioning them lack the experience to assess them.
• The pipeline that produces experienced senior talent is narrowing and nobody is solving it.
• As senior people exit, the knowledge they carry leaves with them. The people behind them never had the chance to build it.
• The answer is not to resist the tools. It is to develop the judgment required to direct them well.
• Nobody controls the swell. The question is whether you know how to move with it.

About the author:

Fynn is Founder and Studio Director of FoSho STUDIOS. He began his career studying video production, filmmaking, and screenwriting, before freelancing across the TV and film industry as a videographer. In 2017 he founded FoSho, originally a video production company, which has since evolved into a partner-led content agency working across science, technology and innovation, public sector, education and social impact, tourism and hospitality, and consumer brands. FoSho STUDIOS is led by Fynn alongside co-founder and Head of Production Martin.

The wave is already moving

We talk a lot about what AI can do. We talk far less about what it is quietly taking away.

Not jobs. Not yet, at least not at the scale most people fear. What it is taking away is something harder to name. The experience of doing the work. The mistakes made in low-stakes environments. The accumulated judgment that comes from years of doing things the hard way before you get to do them the easy way.

That judgment is what separates someone who can direct and evaluate AI output from someone who simply accepts it. And, the mechanism that built it, entry level work, is disappearing.

The swell has arrived. It does not ask permission. It does not move in a direction you choose. And fighting it will not keep you afloat. But before you decide how to move with it, there is a question worth sitting with. Not about the tools. About the people using them.

The grunt work was never just grunt work

I have done real grunt work. Entry level jobs across delivery driving, construction, call centres, hospitality, tourism and more. Jobs across different industries and countries, most with no obvious connection to running a content studio.

But in every one of those jobs I was developing the same skills. Communication. Critical thinking. How to read a situation. How to deal with it when things are not going to plan. The experience was accumulating whether I recognised it or not.

Early in my career working as a lowly Runner, I was instructed by a senior member of a TV production to move a vehicle on set. In a previous job I had been asked to organise insurance for vehicles. So I knew immediately I was not insured to move that long wheelbase van. It was not easy to say I can’t – which they heard a ‘no’. I was able to explain why and I helped avoid a potentially high-risk scenario for me, and the production company. It was not the answer they wanted, but it was the answer they needed. They found someone within minutes who was insured. The situation was avoided.

That judgment came entirely from unglamorous, administrative work in a previous role. It is precisely what an AI system might also miss. Not because it lacks information, but because applying the right knowledge to an ambiguous real-world situation requires contextual judgment that comes from experience, not training data. AI night get there, but it’s not there yet.

This raises a question that nobody has fully answered yet. Professional indemnity insurance exists because someone has to be accountable when decisions cause loss or damage. Right now that someone is a person. As work splits between humans and AI agents, we will need to insure both. And as robotics enters the picture, the economics become even more complex. Will it be cheaper to insure a human doing the work, or an AI-powered system doing the same work? Where will we place our value? We do not know yet. But the question is coming faster than most organisations are prepared for.

Nobody is solving the pipeline problem, but somebody has to

Entry level roles exist not just to get work done cheaply. They exist to bring people into a discipline and give them the experience they need to develop judgment. Remove those roles and you break the pipeline that produces the senior talent every organisation depends on a generation later.

Anthropic’s research, published in March 2026, found suggestive evidence that hiring of workers aged 22 to 25 into AI-exposed occupations has already begun to slow. The mass displacement has not arrived. But the narrowing of the entry point has. The pipeline is beginning to close.

This image is from Anthropic’s ‘Labor market impacts of AI: A new measure and early evidence’, published in March 2026.

The blue area: share of job tasks LLMs could theoretically perform.

The red area: observed real-world usage today. The gap between them will close.

What that chart shows is not just where AI is today. It is the direction of travel. Every sector in the blue that is not yet covered in red is a wave that has not fully arrived. It will.

The more we work with junior members of our own team and with young professionals at the brands we partner with, the more concerned we are about how AI is being used. Not by people with deep domain expertise who can evaluate what comes back. But by people without the experience to know when the output is wrong, incomplete, or inappropriate for the context.

The risk is not that junior staff are doing too much. It is that they are being asked to press buttons without developing the understanding to know whether what comes out is right.

At FoSho we take on work experience placements. Our principle is straightforward: unpaid work experience should never involve work you would otherwise pay someone to do. The placements we value most are driven by broad curiosity, communication skills, and a focus on quality. Not technical ability. Not familiarity with the latest tools. The human stuff. Because that is what you cannot load into an agent.

The kind of person we are looking for now, at every level, is not primarily technically advanced. What we are looking for is someone who is humanly advanced. Communication. Critical thinking. The ability to read a room and make a judgment call under pressure. Qualities built through experience, not downloaded from a course.

We have seen what happens when an industry disappears faster than people can adapt. The mining towns of the UK did not just lose jobs. They lost identity. When the work went, something harder to name went with it. The consequences played out across generations.

The displacement of cognitive labour could repeat that at a scale and speed we have not seen before. This is solvable. It has to be.

The difference is what you bring to the water

A retired musician we know is currently developing an animated children’s TV show. No production background. No budget for writers, designers, or animators. AI has reduced the cost of those contributions to near zero, and a project that would never have existed is now becoming real. That is worth acknowledging honestly.

But we would be dishonest if we did not sit with what it costs.

At FoSho, we are not making fixed decisions about how we absorb AI tools. The development is fast and the direction changes week by week. What we are thinking carefully about is how agents can be introduced into our workflows while maintaining the safeguards, the craft, and the client trust we have spent years building.

The difference between someone who rides the swell and someone who is overwhelmed by it is not the wave. It is what they learned before they got in the water.

Everyone has access to these tools. The tools are not the difference maker. What you bring to them is. The experience. The judgment. The ability to know when something is wrong, when something is missing, and when to say no.

That is not something you can prompt into existence. It is built through years of doing the work. And right now, quietly and without much fanfare, we are removing the conditions that make that possible.

If this resonates, it is worth a conversation.

The questions this piece raises around talent development, institutional knowledge, and how your team evaluates AI-assisted content are ones we work through with our clients every day. If you are responsible for content, communications, or brand and you are starting to feel the shape of these challenges in your own organisation, we would be glad to talk.

Get in touch at hello@foshostudios.com or book a discovery call at foshostudios.com/discovery-call/