Opinion: Engineering AI systems – too crucial to leave to chance?

The increasing importance of artificial intelligence (AI), machine learning and big data in driving economic growth and generating both societal and environmental benefits is recognised by governments around the globe. Prof. Bashir Al-Hashimi, Dean of Engineering & Physical Sciences at University of Southampton and Trustee of the UK Electronics Skills Foundation explores why partnerships between academia, business and the engineering professional institutes are essential to nurturing the talent of the future and establishing the UK as a global leader in engineering AI systems and digital engineering.

The UK Government has already committed to making this country a leader in the use of AI through its announcement last year of the billion-pound AI sector deal, which aims to ensure the UK is at the forefront of the AI and data-driven economy and has the required digital infrastructure and skilled workforce. In light of this, now is an opportune time for those of us in academia to consider the range of graduate skills required to achieve it.

AI systems are much broader than machine learning algorithms. They combine software with sophisticated electronics, pervasive connectivity, machines and physical infrastructure in order to sense, understand, act and, crucially, learn to do things better. These new AI systems envision machines designed specifically to work more cohesively with humans – employing real-time data, adapting to enhance performance and aiding user experience, whilst also creating cheaper designs that consume fewer natural resources.

 

As engineers, we welcome these developments. They offer huge opportunities for engineers as well as computer scientists (and probably almost all other professions) and we need to engage strongly with them. In relation to research, the blurring of traditional boundaries between the distinct disciplines of computer science on the one hand and the engineering disciplines on the other should prove very exciting for the future of AI. Such interdisciplinary work can often lead to remarkable breakthroughs.

However, do we currently have sufficient numbers of engineers engaging with AI? UCAS data (2018) records the total number of UK-domiciled undergraduate students accepted to study electrical, electronic, civil, and mechanical engineering at UK universities as 13,135, significantly fewer than the 15,430 UK-domiciled students accepted to study undergraduate computer science degrees.

It is imperative that we continue to train computer scientists but equally, we also need engineers with an understanding of AI. The trend of student study choices towards computer science seems likely to continue over coming years, fostered by extensive media coverage of AI’s role in the economy, its identification with computer science and software, the significant investment in computing at schools without a matched focus on engineering, and the impact on every aspect of society.

However, in our ever more connected and automated world, next-generation AI hardware will need to be more powerful, more reliable and more cost-efficient. This has the potential to transform engineering and I believe that the engineering curriculum in UK universities needs to immediately embrace AI, machine learning and data science in a more coordinated way in teaching engineering design and practice.

 

I believe passionately that we must take action now to ensure that our future programmes provide the people and the expertise to lead the engineering design, management and training of the AI systems of the future – systems that will underpin autonomous transportation, intelligent large-scale infrastructures, and smart personalised healthcare. Indeed, engineering AI systems is a collaborative process and cannot be achieved solely by academics working on their own. Strong partnerships between academia, business and the engineering professional institutes are essential to ensure the design and relevance of these new courses to the existing and emergent industries for which we urgently need to train our graduates.

It is already estimated that the UK needs to produce an extra 20,000 graduate engineers every year in order to sustain the industry, and it is predicted that a further 260,000 skilled people are needed if the UK is to meet its ambitions to invest 2.4% of GDP in R&D by 2027 in making the UK the most innovative country in the world. As engineers, we have a strong obligation to establish the UK as a global leader in the AI systems field and to ensure this exciting opportunity fulfils its promise.