How Is AI Impacting Central Government?

The AI Council was established in 2019 to offer advice and information to the UK’s Government, and to support the work that the AI Sector Deal had already outlined. Two years later, the Council published its AI Roadmap which provided recommendations to help the Government decide its strategic AI direction. Here we look at the Government’s strategy, what’s been adopted within central government, and the issues surrounding it.

The UK’s National Artificial Intelligence Strategy

The National AI Strategy has set its aims as:

• Investing and planning for the long term needs of AI in order to continue the UK’s ‘superpower’ status within science and AI technology

• Supporting and transitioning the country to an AI-enabled economy to benefit all sectors and regions

• Ensuring the UK gets AI governance correct to encourage innovation and investment, and to protect the public and our ‘values’

The way to achieve these goals, says the strategy, is to engender public trust and support, and to involve diverse views and talents. 

In January 2021 the AI Council published its AI Roadmap, with 16 recommendations for the strategic direction of AI. These include: 

Research and Development – including AI investment, accessing top talent from around the world; finding new ways of bringing researchers, disciplines and sectors together; elevating the Alan Turing Institute to a national institute and providing it with long-term public funding.

Skills and Diversity – including scaling up and committing to a 10-year programme of AI skills building, including research fellowships, AI-relevant PhDs, industry-led Masters, and Level 7 apprenticeships; making diversity and inclusion a priority to ensure that underrepresented groups are given equal opportunities in all programs; committing to achieving AI and data literacy so that everyone can understand the risks and rewards of AI, including an online AI academy to support teachers, school students and lifelong learners.

Data, Infrastructure and Public Trust – including consolidating and accelerating the infrastructure that AI needs; ensuring safe data sharing; leading the development of data governance; ensuring public trust through public scrutiny; make the UK a global leader in good governance, standards and frameworks for AI and enhancing cooperation with other major AI nations.

National, Cross-sector Adoption – including increasing buyer confidence and AI capability across all sectors and sizes of company; supporting the UK’s AI startup vendor community to enable greater access to data, infrastructure, skills, specialist knowledge and funds; investing in public sector AI; using AI to meet Net Zero carbon emission challenges; using AI to keep the country secure; and building on the work of NHSX to create value in healthcare.

What AI is already in use?

The Government wants to make more use of AI to benefit the public sector. In particular, in a 2020 publication entitled, ‘A Guide to using AI in the public sector’, it highlights the potential AI has to provide more accurate medical diagnoses, solve social problems, allow politicians to experiment with policies before committing to them, improve public services, and automate time-consuming tasks to free up frontline workers. 

Let’s take a look at a couple of real-world examples already in use:

Machine Learning making Gov.uk more accessible

The Government Digital Service wanted to make information on its website more accessible to its users. Originally it wanted to organise information more locally but found that tagging 100,000 untagged pages was time-consuming and costly. Using a combination of classification and Natural Language Processing (NPL) it was able to build a supervised machine learning model to make text content on the page machine-readable. These results were then used to learn patterns which could predict where the untagged pages would fit, and was able to provide tags for 96% of the existing content, as well as suggest tags for new content. It was accomplished within 6 months. 

Colouring London

Colouring London, developed by the Alan Turing Institute, is helping with the UK’s Net Zero ambitions and trying to make it more sustainable by providing data on the types of building that London is composed of, how many building there are, their energy efficiency, their age, size and location, as well as who owns them and whether they work within the context of a local community. The project looks to develop knowledge sharing with permanently open databases which will be maintained collaboratively, with information added by members of the public, academic bodies, Central Government, industry and the not-for-profit sector. 

This week saw two items of news about AI: firstly that Geoffrey Hinton, Google’s AI pioneer resigned citing the ‘dangers’ of AI: and secondly that Britain’s Competition Regulator announced that it would begin to examine the impact of AI on society, and whether new controls were needed. These issues raise more questions than they answer in terms of the development and future of AI. Whether regulation is needed? Whether our data is secure? Whether eventually AI can do more harm than good. What’s clear is that this new and emerging technology must be handled with respect and caution by those who seek to utilise it until its full potential can be assessed.