Dr Clare Walsh from the Institute of Analytics examines the educational challenges in AI transformation and questions whether the UK’s AI strategy is effective
Upskilling an entire nation to be data literate is a challenge. Decision-making is going digital, though, and it can’t be ignored. Digital decision-making is just as vital to the economy as digital communications. The government’s AI strategy is an opportunity to support digital growth through skills, adoption and investment in infrastructure.
In terms of nurturing and attracting talent at the highest levels, the government-funded places for MSc conversion courses that were introduced in 2020 boosted the provision of data training. Large cohorts now take advantage of a healthy MSc provision for specialised data science training in the UK (UK Office for Students, 2023).
An AI strategy: Educational challenges in AI transformation
The UK’s main private AI research organisation, DeepMind, requires a PhD as a minimal job entry qualification. Public funding focused on attracting top talent to PhD programs will support research efforts in the UK. It may attract globally strategic AI companies to expand their footprint in the UK further. However, career support for PhD graduates to stay in the UK in private sector employment remains disappointing, and we will continue to lose talent to countries that assume top talent will work in the private sector.
Training for all is also important. Adoption continues to lag behind technical development. The predicted revolution in 2013, where 50% of roles would be performed by machines by 2023, has not come to fruition (Frey and Osborne, 2023), mainly because these socio-technical machines require user buy-in. Businesses and industries must be ready to embrace them, requiring the wider population to be minimally informed on data processes.
As we undergo transformation through technologies founded on mathematical reasoning, weaknesses in mathematics provision, especially in England, remain under the spotlight. Optional mathematics provision at the upper secondary level makes England a global outlier. Among other nations, we are still far below mathematics standards at 16-18. Working with many AI tools requires some understanding of numerical reasoning and the interpretation of statistical reporting. Business and industry will continue to be expected to make up this shortfall in education or recruit internationally.
Measuring AI readiness in the broader population
Measuring AI readiness in the broader population will be difficult. Decades on, we still have no one standard of measuring digital communications standards in the population. We can count graduates from dedicated data science programs, but data is a vast and complex field with multiple dependencies and infrastructure requirements. Reliable and valid measurement of these skills presents challenges. A low understanding of the skills required to work with AI may hamper some of the other objectives of AI policy.
For example, the provision of open data continues the efforts of successive UK governments to create an interoperable data ecosystem across the UK. Academics see the benefits clearly. It will allow startups, higher education and more established businesses to thrive. However, it comes with complex legal frameworks and standardisation expectations that have always been challenging to communicate to a busy working population (Priestley, Simperl, Juc and Anguiano, 2022). It will be easier to sell public funding that is aligned with short-term goals that are readily comprehensible.
One thing missing that may enable the training goals of the AI strategy is the provision of a standardised certification in AI readiness. It may encourage private sector development of AI training and skills measurement. It would also provide the government with a means to measure the effectiveness of public spending on training with a view to impacting adoption.
