Comment: STEM education must rise to AI challenge

Full disclosure: I used multiple AI tools to help me write this. To paraphrase Isambard Kingdom Brunel: ‘The educated are not those who know the answers but those who know where to find them.’ Understanding which tools are best for a task and how to use them effectively has always been central to the work of a professional engineer. However, with an education system shedding teaching staff and struggling to recruit at target levels - the Teacher Labour Market in England Annual Report 2024 revealed that just 17% of the required number of physics teachers started training in 2023 (with the picture likely similar in the rest of the UK) - the STEM education delivery is clearly not fit for purpose. It should ensure all school leavers can identify, understand, and harness state-of-the-art innovation. In a technology-driven society, these skills are no longer a luxury but absolutely vital.

So, what are the problems to be solved? Recruitment of teachers with technical skills continues to be a serious challenge. Despite this, the government has lowered the 2024-2025 postgraduate secondary STEM trainee recruitment target. Not that the objectives seem to matter, having been met just once in the past decade.
 
Having teachers in post is only part of the solution. With technology changing so quickly, continuing professional development (CPD) in digital competencies should be widespread. Alarmingly, the Pearson 2023 School Report notes that only 17% of educators were receiving training on digital tools and edtech advances, while the Digital Poverty Alliance’s Tech4Teachers White Paper indicates that a lack of teacher capacity is a major driver of this gap. It is hardly surprising: pupil-to-teacher ratios have risen from 14.8 in 2010/11 to 16.8 in 2023/24 for state secondary schools in England. Equal access is also an issue. According to research from Oriel Square, teachers in private schools were found to be twice as likely to have used AI for schoolwork than those in state schools. Getting more STEM teachers into schools and giving all teachers access to high-quality CPD - or other forms of ongoing upskilling - is essential to preparing the next generation for careers shaped by advances in artificial intelligence.
 
STEM charity EngineeringUK recommends that the government reverses recent cuts to existing CPD programmes where funding allocations for CPD for science teachers are set to drop by almost half between the 2023/24 and 2024/25 financial years, from £8.4m to £4.5m. EngineeringUK has also strongly advised the government to invest a further £29 million per annum in high-quality CPD for STEM teachers, as it is proven to yield a significant return on investment through increased retention rates and improved quality of STEM teaching.
 
Of course, there is uncertainty about exactly what the future world of work will look like. Research recently commissioned by Morrisby outlines five future scenarios, ranging from a gradual, incremental expansion of AI to rapid, widespread, and highly disruptive technological gains. Whichever path emerges, it is clear that education needs to change, so the government’s Curriculum and Assessment Review is a welcome opportunity to try to address this. Government must not miss this opportunity and ensure that we have a curriculum fit for generation AI.

MORE FROM AI WEEK
 
While most AI-related skills mirror those needed for other technologies, certain areas deserve particular focus. Building AI tools requires integrating maths, statistics, data science, and computer science. The Royal Society Mathematical Futures Programme makes a compelling case for a three-part structure - foundational and advanced mathematics, general quantitative literacy (GQL), and domain-specific competencies (DSC) - to deliver exactly that. Arguably, for maximum impact, it should be implemented across all age groups to open diverse career paths for all students. The old strategy of replacing formal ICT teaching with digital skills across the curriculum must shift to a dual approach, in which DSC are taught widely in both vocational and non-vocational subjects.
 
How we assess students also needs to evolve. Oxford, Cambridge and RSA Examinations (OCR) has called for the government to introduce a “short course” maths qualification to be taken by all students in the first year of their GCSEs, which feels like a step in the right direction. Some computer science GCSEs currently require no actual use of a computer, favouring written exams. Furthermore, Design and Technology burdens pupils with 50% written assessment. This combination is both off-putting to many learners who enjoy the practical elements of these subjects and poor preparation for the careers of today, never mind those of tomorrow.
 
As an engineer who has written software in one way or another for most of my career, the recent advances in language models have already transformed my daily work. Ideas that, five years ago, I would have abandoned - because days of trawling through documentation and online Q&A sites were unlikely to be worth the payoff - can now be evaluated in a few hours.
 
We must create an education system that channels the creativity and passion of young people into rewarding careers while teaching them about advances in science and technology. Otherwise, who knows how many great ideas we will be missing out on.
 
 

 Steve Rossiter is a Technology Consultant and EngineeringUK Trustee