Here is a question nobody in the room wants to ask when we talk about AI and the future of work. I’m not talking about the question of which jobs will disappear, or what skills you need to stay relevant, or how to position yourself in a market that is changing faster than most people can track. Those are real questions, but they are downstream of a more fundamental one, the one that the last hundred years of industrial labour have been carefully avoiding: what are human beings actually for?

Not in terms of output or productivity or GDP contribution. What are we, biologically, neurologically, at the level of what our species actually evolved to do, and how viciously has the modern economy misaligned us from that?

The reason this question matters right now, specifically, is that AI makes the misalignment undeniable. For the first time, we can look at what machines are genuinely good at and hold it up against what humans are genuinely good at, and the comparison is not flattering to the economy we built. Not because humans lose, because we lose at exactly the things the economy valued most, speed, volume, consistency, availability, and the things machines cannot touch are the things we spent fifty years being told did not count as work.

The human brain runs on approximately 20 watts of power. Your phone charger uses more electricity than the most sophisticated biological computing system in the known universe. That brain contains roughly 86 billion neurons, connected by an estimated 100 trillion synaptic connections, operating in parallel, continuously, from before you are born until the moment you die. It rewires itself based on experience. It builds models of other minds. It generates meaning from incomplete information. It feels.

According to research from the Blue Brain Project at EPFL in Switzerland, simulating the full processing capacity of a human brain on current AI infrastructure would require approximately 2.7 billion watts. That is not a rounding error. That is a gap of 135 million to one, to fully simulate the brain's processing capacity. The most energy-efficient organ ever produced by four billion years of evolution is running laps around the most advanced computing infrastructure our species has ever buil, and doing it on the energy equivalent of a dim light bulb.

This is not a feel-good statistic. It is a precise technical measurement of the gap between biological and artificial intelligence. And it points to something important: the human brain is not slow, not inefficient, not a problem waiting to be solved by better hardware. It is an engineering marvel that current technology cannot come close to replicating, and the economy has been treating it like a component to be optimised out.

In March 2025, MIT Sloan published research building what they called the EPOCH framework, five categories of human capability that AI fundamentally cannot replicate. Not "not yet." Not "in current models." Cannot. The researchers were direct: these are not gaps waiting to be closed by the next model release. They are structural limitations of what artificial intelligence is.

Look at that comparison carefully. The things AI does well are things the economy has spent a century trying to make humans do: consistent, available, high-volume, fast, tireless. The things humans do that AI cannot replicate are the things that were historically dismissed as soft, immeasurable, unbillable, or irrelevant to productivity. The economy got the wrong column. For a hundred years, it optimised for the column that machines were eventually going to own anyway, and it did so at the cost of the column that no machine can touch.

"The economy optimised for the wrong column. For a hundred years it tried to make humans behave like machines — and then built machines. The things that got left behind are the only things that cannot be replaced."

This is not an abstract philosophical point. There is a body count.

In 2025, research found that 82% of employees across North America, Asia, and Europe reported being burned out. Not slightly stressed. Burned out, a clinical state characterised by exhaustion, depersonalisation, and a reduced sense of personal accomplishment. A 2025 study of Korean workers found that burnout mediated 51% of the total effect of occupational stress on health-related productivity loss. Burned-out employees have presenteeism rates 4.7 times higher than their colleagues, they are physically present and mentally gone, because the job asked them to be a machine and their nervous system eventually stopped cooperating.

Gen Z and millennial workers are hitting peak burnout at 25. Twenty-five years old. We have built an economy so aggressively misaligned with human biology that it is burning people out before they have had time to figure out who they are.

_{82% of global knowledge workers reported burnout in 2024–2025  ·  DHR Global / Skillsoft research}

Burnout has been clinically linked to a 21% increase in cardiovascular disease risk, increased stroke risk, depression, and an 84% increased risk of Type 2 diabetes. This is not a productivity problem. This is a public health crisis produced by asking biological organisms to behave like systems they are not. And we keep doing it. We keep writing performance frameworks that reward availability and penalise boundaries. We keep building cultures that mistake exhaustion for commitment. We keep making children watch their parents disappear into work and then presenting that disappearance as success.

Every major technological transition in human history has been accompanied by the same choice: does this technology serve human life, or do human lives serve the technology? The industrial revolution could have reduced working hours as machinery absorbed repetitive labour. Instead, it extended them. The productivity gains went to capital. The cost was paid in bodies.

The digital revolution could have freed knowledge workers from administrative repetition. Instead it created always-on availability, the colonisation of personal time by work communication, and the expectation of a response at any hour because the technology made it physically possible to send one. The productivity gains, again, went upward. The cost, again, was paid by the people doing the work.

We are at the third inflection point. AI is absorbing, genuinely, rapidly, usefully, the repetitive, high-volume, pattern-recognition work that the economy has been asking humans to perform at the cost of their health and their humanity. This is an opportunity.

The question is not whether AI will take your job. The question is whether you are willing to build a life around the things that no machine can do — the judgment, the presence, the creativity, the connection — and whether the economies and institutions we are building are going to make space for that, or whether we are going to let the productivity gains flow upward again while the next generation burns out at twenty-five instead of forty-two.

This is not a romantic argument for going off-grid. It is not a manifesto against technology. It is a reading of the evidence, the neuroscience, the labour research, the energy data, the MIT frameworks, and a statement of what that evidence plainly implies: the thing that makes human beings worth anything to each other, and the thing that no amount of compute can replicate, is our subjectivity. Our capacity for meaning. Our ability to be present to another person's experience. Our creativity, which is not pattern recombination but the genuine expression of a consciousness that has lived in a body in a world.

If you are building a freelance practice, a creative business, a body of work, you are doing something that requires all of the left column and none of the right one. You are not competing with AI on pattern recognition or output volume. You are doing the thing that the right column cannot do: being a person who has something to say, to someone who needs to hear it, in a way that lands because you are real and they know it.

The data on what humans are, at the level of the 20-watt miracle of tissue and electricity in your skull, is not a consolation prize. It is the description of the most sophisticated system in the known universe, one that has spent four billion years evolving capabilities that we are only just beginning to map, and that a trillion dollars of compute cannot come close to replacing.

Current status: humans doing machine work. Machines doing human work. Nobody questioning this. Carry on.

Sources: MIT Sloan EPOCH Framework research, Rigobon & Loaiza (March 2025); University of Amsterdam / PNAS, "Affordances in the brain" (June 2025); Blue Brain Project, EPFL; Texas A&M / Science Advances super-Turing AI energy research (2025); Nature Human Behaviour, Luo et al., "Large language models surpass human experts in predicting neuroscience results" (2025); DHR Global burnout survey (2024–2025, n=1,500); Emory Economics Review, "The Economics of Burnout" (2026); Kim et al., Industrial Health burnout productivity study (2025); Growthalista burnout meta-analysis (2025); Frontiers in Psychology, "The Illusion of Empathy" (June 2025).

Subscribe here
Visit Profreelance