Education – And Its Limits

Education – and its Limits

Daniel Susskind, Gresham Professor of Business

26th May 2026

We live at a remarkable technological moment. Almost every day, we hear of new technologies that are taking on tasks and activities which, until recently, only human beings alone could ever do. The recent progress in generative AI – systems like ChatGPT, Claude, Gemini, and Grok – are remarkable. But the argument of the lectures until now is that these are a remarkable chapter in a far longer story. This burst of technological progress has not come out the blue. Nor is this the first time our societies have had to find an effective response to technological disruption. 

As we have seen in previous lectures, the main response to technological progress in the 20th century was ‘more education’. The labour market challenge was often thought of as a ‘race’ – new technologies were racing ahead, and workers needed to be given the skills that would allow them to keep up. 

However, what the ‘more’ in ‘more education’ meant changed a great deal over time. At the start of the 20th century, ‘more’ meant more people. This was slow in coming. In the 1930s, as the economists Lawrence Katz and Claudia Goldin noted, the US was “virtually alone” in providing free secondary school education. But as time passed, other countries caught up and copied that initiative. Today, that sort of education is commonplace. 

But by the end of the twentieth century, the meaning of ‘more’ had transformed. It no longer meant simply more people, increasing access to schooling for all, but more advanced education, with a focus on encouraging people to attend college and universities. At the turn of the century, it was common to hear political leaders talk about the importance of this sort of education. 

In this lecture, I will argue that ‘more education’ remains our best response to technological disruption in the world of work. As I have explored in previous lectures, it is likely that the main challenge in the labour market for now from a technological point of view – and in the medium term – is not that there will not be enough jobs. Instead, there will be jobs – but for various reasons, these jobs may sit out of reach of those workers who want them. Put more formally, in the language of this series, the challenge is likely to be ‘frictional’ rather than ‘structural’ technological unemployment. And one of the best responses we have to the former remains ‘more education’. 

However, just as the meaning of ‘more education’ has changed in the past as the nature of technological progress has changed, it will have to change again. In this lecture, I will argue that uncertainty about the future – about exactly which jobs will have to be done, about exactly what skills will be in demand – make it very difficult to proceed as we have in the past. In particular, the idea that we can pick out ‘future-proof’ skills that are out of reach of AI and teach people to do them is unlikely to succeed. In short, we simply do not know enough about what lies ahead for that to be effective.  

Instead, I want to explore what we ought to do – in particular, how we ought to change what we teach, how we teach, and when we teach – to prepare people for this uncertainty and prepare people to flourish in the working world that is emerging. 

However, in closing, I also want to explore the limits to education. Too often – in part due to its historical success – ‘more education’ is treated as a panacea. This is a mistake. To begin with, it is an imperfect response to the challenge of frictional technological unemployment – as we have seen in previous lectures, ‘skills’ are only one reason that people might struggle to take up available work. In turn, it is an ineffective response to the challenge of structural technological unemployment – if we find ourselves in a world where there is not enough work to be done full-stop, we will have to consider very different responses. 

Further Reading

‘An OpenAI model has disproved a central conjecture in discrete geometry’, OpenAI.com, 20 May 2026. 

Bloom, Benjamin. ‘The 2 Sigma Problem: The Search for Methods of Group Instruction as Effective as One-to-One Tutoring’, Educational Researcher. 13:6 (1984). 

Deming, David et al. ‘Navigating Public Job Training’, Harvard Project on Workforce. March 2023. 

Katz, Lawrence and Claudia Goldin. The Race Between Education and Technology. (HUP, 2008).

McBain, Sophie. ‘Are we living in a golden age of stupidity?’, The Guardian. 18 October 2025. 

Susskind, Daniel. A World Without Work. 2020. 

Susskind, Daniel. ‘Technological unemployment’, in The Oxford Handbook of Economic Governance ed. Justin Bullock et al. (2022). 

Susskind, Daniel. The Digital Papers: The Economics of Transformative AI. (Stanford Digital Economy Lab, 2025). 

Susskind, Daniel. What Should My Children Do? (Allen Lane, September 2026).

© Professor Daniel Susskind 2026