AI will kill off the lingering myth of meritocracy
We've gaslighted ourselves for years into believing that people's life trajectories owe much more to merit than luck. By removing our ability to measure merit, AI will finally put paid to that idea.
A year ago, I wrote a piece for the RSA about myths of merit in an unequal society, in which I argued that the meritocratic narrative both masks and justifies economic inequality. What looks like merit is mostly luck: the luck of being born to the right parents, in the right place, with the right start. If we paid more attention to how much our life trajectories are influenced by this broad conception of luck, we would take bolder action as a society to reduce the inequalities that underpin it.
In the intervening twelve months, the capabilities of large language models have developed enormously. I don’t know about you, but I have shifted rapidly, and recently, from being sceptical about the impacts of AI to being pretty convinced that it will drastically reconfigure our economy over the next few - and I mean few - years. I’m far from alone in thinking this (although I do think that many people have their heads in the sand), and the British public is also concerned. A few recent articles worth reading on this include those by Rutger Bregman, Martin Wolf in the FT, Simon Johnson (chair of the UK’s new AI Economics Institute), Roa Powell, Matthew Harvey Sanders, and Pope Leo XIV.
And some of the projections are sobering. Sanders quotes Mustafa Suleyman, CEO of Microsoft AI, predicting that most white-collar work will be “fully automated by an AI within the next twelve to eighteen months.” He cites estimates of a 50% chance that artificial general intelligence is with us by 2030 - within four years - leading to one person doing the work of hundreds. And he quotes Demis Hassabis, CEO of Google DeepMind, who said last summer that “AI will be ten times bigger than the Industrial Revolution and perhaps ten times faster, unfolding over a decade rather than a century”.
We’re not paying nearly enough attention to the tsunami that is headed our way, and to the drastic steps that we need to take now to prepare for it. As Martin Wolf said in the FT, “there is a good chance that AI will in time devastate the labour market, increase inequality and create an extraordinary concentration of economic - and so political - power in the hands of a tiny number of businesses and people… A good part of the increased income and wealth must be shared. The time to prepare for this is now. If we do not act, it will be far too late”.
However, in this post I want to focus on one even more neglected aspect, that has important implications for how we think about ourselves and each other. If the myth of meritocracy is already wobbling on shaky foundations, I suspect that the AI revolution will finally shatter it - with implications both good and bad.
The existing critique of meritocracy largely rests on the fact that life chances and outcomes owe as much to luck as to merit (talent + effort). This is exacerbated by the fact that both our society and the labour market values cognitive skills more than manual labour or care work, as David Goodhart argued in Head Hand Heart. But the rapid adoption of AI is completely undermining our ability to measure cognitive talent and effort in the first place.
Before AI, we would look at what someone produces, whether it was a report, a strategy, a diagnosis, or an elegant piece of code, and would be able to infer fairly accurately the level of both effort and talent that must have gone into it. In theory, the person who produced this output would be rewarded accordingly - with money, but also with status, respect, and the unspoken assumption that these rewards were fully deserved.
Critics of meritocracy, including myself, would argue that the ‘inputs’ into this process were unequally shared out, and were not ‘earned’. People were born with different talents, into families with different amounts of wealth, in poorer or richer parts of the country, into very different social networks, and so on. So the outputs could not be judged against a fairly distributed set of inputs.
However, what was at least relatively safe (bar cheating) in the pre-AI era was the assumption that the outputs belonged to the person. The fairness was upstream of the task. This is no longer the case. When anyone with a monthly subscription to an LLM can produce a competent strategy, marketing plan or web prototype in minutes, it is no longer possible to judge the ‘merit’, however deserved or not, of the person who submitted it. The signal that work normally sends about the worker has dissolved into noise. The myth of meritocracy depended on legible proof of desert, but AI has made the proof illegible.
Luck undermined meritocracy because the race was rigged, but at least the times on the clock were real. AI destroys meritocracy because we can no longer trust the results of the race at all.
Some might read this as an uncomplicated good news story, that capability is democratised, that everyone’s talent is unlocked. And that might be true to an extent in a future economy in which AI is mostly used to augment workers. But, as the articles cited above suggest, the more likely scenario is the rapid use of AI to replace workers. And in this scenario, an increasing share of the rewards goes not to the individuals who nominally produce the work, but to the people and organisations who own the model that did much of the actual work, the data it was trained on, the compute that it runs on, and the platform that distributes it. Reward is detached from merit, and from labour, and re-attached to ownership of the infrastructure.
Unfortunately, this is a trend that has been accelerating in our economy over recent decades anyway, quite apart from the impacts of AI. As Piketty observed in 2014, capital is attracting higher returns than labour, in a self-reinforcing cycle. Increasing amounts of wealth are extracted rather than created, in particular by companies who own a scarce asset and charge excessive amounts of ‘rent’ to access it, from housing to land and infrastructure. AI looks poised to supercharge our already-advanced transition to a rentier economy, except that this time the rent is also being extracted from cognitive labour. Talented and hardworking knowledge workers face being rapidly ejected from the top of the meritocratic pyramid, and being turned into tenants, digital serfs, paying for the tools that do the real thinking while the companies that own the real means of production reap the rewards.
To those who object that technological advances have always changed working habits - that the calculator did not undermine mathematical talent, that the internet did not destroy the job market for researchers - the AI revolution looks different. Unlike its predecessors, it doesn’t just automate the periphery of knowledge work; instead, it can take over the core components, the thinking, the reasoning, the writing. This reduces the role of human judgement to knowing what to ask and how to check the answer - hardly the basis for a functioning system of reward based on meritocracy.
If we collectively recognised this state of affairs, it could unlock a consensus for making some of the radical changes to our society and economy that the AI era demands. If we can no longer credibly tell people that their place in society reflects their merit, we need a different basis for deciding how both economic resources and intangibles such as social status are distributed. We cannot keep on distributing money and respect based on a metric that has no meaning. And we need to recognise that the coming ruptures to the labour market will threaten not only people’s livelihoods but also their sense of purpose and meaning, unless we take radical steps to rethink how we organise our society.
That means taking seriously the ideas that the myth of meritocracy was designed to keep off the table - reducing inequality, distributing resources more fairly at source rather than relying on the tax system to do what it can to redistribute them, spreading ownership of tech stocks so that its returns are widely shared rather than privately captured, and decoupling human dignity from a merit metric that no longer works. It means, ultimately, a new social contract, grounded not in the fiction that we get what we deserve, but in the question of what we owe each other. The myth of meritocracy allowed us to spend decades collecting gaslighting ourselves into believing that this question was already answered, but it never was. Now we no longer have the excuse of pretending otherwise.