The hallowed halls of Stanford University have long been a beacon of meritocratic aspiration, a place where raw talent and hard work could earn a golden ticket to the American dream. But a new controversy threatens to tarnish that image, with UK university leaders warning that the use of artificial intelligence in elite admissions could be rigging the game in favour of those who can afford to play it.
The debate was sparked by revelations that Stanford’s admissions office has been piloting an AI tool designed to sift through thousands of applications, flagging candidates with the highest ‘potential for success’. The algorithm, trained on decades of historical admissions data, promises to reduce human bias and streamline a notoriously opaque process. But critics argue it encodes past inequalities, favouring applicants from wealthy backgrounds who have access to premium test prep, extracurricular coaching and polished personal statements.
In a joint letter to the UK’s Office for Students, the Russell Group of universities warned that such tools could ‘calcify privilege’ if adopted without rigorous oversight. ‘Meritocracy is not a dataset,’ said Professor Alice Thorpe, vice-chancellor of the University of Bristol. ‘An AI trained on who got in before will simply perpetuate the same patterns of advantage. We risk creating a system where the algorithm decides your fate before you’ve even written your personal statement.’
The irony is not lost on Silicon Valley expats like myself, who have watched the tech industry’s love affair with AI slowly bleed into every corner of our lives. Here, in the heart of the innovation economy, we are confronted with a Black Mirror scenario: an algorithm that claims to democratise opportunity but may actually lock in privilege. The technology itself is not new. Predictive analytics have been used by retailers to target ads and by banks to assess credit risk for years. But applying them to education is a different beast, one that touches the very idea of social mobility.
At the core of the controversy is the concept of ‘fairness’. Proponents of AI admissions argue that humans are notoriously bad at predicting success, swayed by unconscious biases around race, gender and class. An algorithm can be trained to ignore these signals, focusing only on ‘objective’ metrics like test scores, grades and leadership roles. But as MIT’s Joy Buolamwini has shown, algorithms are only as unbiased as the data they are fed. If the historical data reflects systemic inequalities, the AI will simply replicate them at scale.
Stanford has been tight-lipped about the specifics of its tool, citing commercial sensitivity. But leaked documents obtained by the Stanford Daily suggest that the algorithm weights ‘extracurricular distinction’ heavily, a factor that correlates strongly with family income. This is not a bug, but a feature of how we define merit in elite circles. A student who spends summers volunteering at a tech incubator in Palo Alto has a very different profile from one who works part-time at a supermarket in Hull. Both may be equally brilliant, but only one fits the algorithm’s mould.
The UK’s higher education sector is watching closely. With the rise of foundation years and contextual offers, British universities have been trying to level the playing field for students from disadvantaged backgrounds. An AI that undermines these efforts would be a serious blow. ‘We are committed to admitting students with the greatest potential, regardless of background,’ said a spokesperson for the University of Oxford. ‘Any use of AI must be transparent and accountable, and must not entrench privilege.’
As a technology and innovation lead, I see this as a watershed moment for digital sovereignty. We must ask who owns these algorithms, who has access to them and whose interests they serve. If we allow AI to become the gatekeeper of elite opportunities, we risk creating a caste system sorted by code. The solution is not to abandon AI, but to design it with fairness baked in from the start. That means diverse training data, ongoing auditing and a commitment to transparency.
For now, the golden ticket remains a lottery. But the stakes could not be higher. If we get this wrong, we will have outsourced one of society’s most important judgements to a machine that cannot understand the human story behind each application. As Professor Thorpe put it: ‘The future of meritocracy depends on our ability to keep the human in the loop.’
