Challenges of implementing artificial intelligence to manage public services more efficiently

The public sector has the potential to become more efficient and productive through the adoption of AI, but rolling out AI across the public sector is a major challenge facing Labour at the moment. As the NAO report authors warned, large amounts of good quality data are important for training, testing and deploying AI models. “Our research found that limited access to good quality data was a barrier to AI adoption, and central government support was important to address this,” they said.

Elena Simperl, research director at the Open Data Institute, believes the UK urgently needs to expand the narrative around AI to understand data and the needs of those using it. “AI brings significant opportunities for efficiency, growth and innovation,” she said. “If we want to unlock the potential of the UK’s AI and secure our place as world leaders, we need to put data at the heart of what we do.”

However, Simperl warned that the UK faces a number of challenges, from a shortage of basic data skills to concerns about data reliability and data sharing, all of which she said needed urgent attention if the UK was to fully seize the opportunities AI could provide.

She added that both the public and private sectors are increasingly relying on data and new technologies – including artificial intelligence – to increase efficiency and improve services and products, so everyone should be aware of the opportunities and constraints.

“Differences in access to data and information persist, creating a digital divide that impedes social progress and economic development,” Simperl said. Data from ODI shows that 30% of the UK population is not aware of the most prominent generative AI products, including ChatGPT2. “Those who are familiar with new generative AI tools do not always understand how to use them productively and responsibly,” she said.

Discussing responsible use of AI, Anna Dent, Head of Research at Promising Trouble, said: “Automation relies on huge amounts of data, which raises concerns about data sources and citizen consent. The controversy over the Data Protection and Digital Information Act highlights the delicate balance between data use and privacy rights.”

She urged policymakers to focus on upskilling staff to equip public sector workers with the knowledge to understand and assess automated systems. She recommended that public sector decision-makers examine the underlying reasons for automation to ensure that such systems do not replicate or exacerbate existing problems. Dent said that when public sector bodies consider implementing automated systems, they must be implemented with consistent impact assessments that enable authorities to modify harmful systems.

One possibility – at least in the private sector – has been to use AI to enhance chatbots, which alleviates some of the burden on call centre staff. Looking at chatbots in public sector services, Paul Waller, research director at Thorney Isle Research, pointed to the limitations demonstrated in the Government Digital Service’s experiment with a chatbot interface to Gov.uk.

He said GDS found the chatbot’s responses did not reach the level of accuracy required for a site where factual accuracy is key. “For that reason alone (and the lack of explainability and consistency), they are not suitable for use in statutory administrative procedures,” Weller said.

Waller also urged policymakers not to use such AI tools to summarize policy consultations. “A chatbot summary will not allow for the nuanced positions of stakeholder groups on a policy,” he said. “Furthermore, even if accurate, an automated summary does not fulfill the democratic function of consultation to enable all voices to be heard and to show that they have been heard. Similar issues apply to using these tools to generate policy advice.”

It is clear that the new Labor government needs to have an open debate about the risks and opportunities of deploying AI and automated systems in public services. Eduardo Ustaran, partner and co-head of privacy and cybersecurity at Hogan Lovells, said AI is a growth opportunity that cannot be dismissed as too dangerous.

“The new UK government should be guided by the universal principle that technological developments are designed to help humanity by increasing prosperity and well-being, so good regulatory policy will actively support technology developments, focusing on promoting good outcomes for all,” he said.