Generic filters
Filter by Categories
Upcoming Series
Filter by content type

A Sluggish Public Sector – An AI Solution


HMRC’s high-handed attempt this week to reduce its telephone service adds to the failures of the public sector imposed on long-suffering taxpayers and the public.  But, says Ross Kempsell, if the potential of AI can be realised, it will bring better government and happier voters.

This week HMRC announced plans to close its customer service telephone line for half the year. Unsurprisingly, the outcry was immediate. Already, according to the Public Accounts Committee, the average waiting time for a call to be put through to HMRC is now 16 minutes and 24 seconds. Six in ten callers wait more than ten minutes for their enquiry to be answered. Many of these come from the 12 million who complete self-assessment tax returns every year and need help: people taking their first steps into business with an extra job, an online ‘side hustle’, the self-employed ‘sole trader’ earning more than £1,000, small business owners, contractors, creators. When I spotted this proposed limitation of public sector customer services, I flicked open my laptop and – with righteous steam pouring from my ears – submitted a written question on the matter in Parliament, seeking a reversal.

Before my question could even be tabled, Jeremy Hunt was swift to intervene – and within 24 hours the decision by HMRC had indeed been quashed (I suppose that when the Chancellor calls HMRC, he is put straight through). This reversion to common sense, however, required a very senior ministerial intervention. So how can we be confident that the government and the wider public sector is truly focused on improving the quality and pace of delivery of the service it gives to taxpayers and citizens, or, as it euphemistically calls them ‘customers’ or ‘service users’?

There are good reasons for the public sector to fully embrace the power of artificial intelligence (AI) tools and technology to achieve more effective delivery. AI tools have the potential to revolutionise our experience of public sector services. But even allowing for some general anxiety about their use, they are still significantly underused. This month’s report from the National Audit Office on ‘the use of artificial intelligence in government’ reveals the woefully weak implementation of AI technologies across central government and its arm’s-length bodies. Apparently, just 37 per cent of the 87 government bodies responding to the NAO survey had deployed AI, with only 74 individual AI use cases deployed across government.  Notwithstanding such low deployment, the opportunities for the use of AI tools in government is broad. The Land Registry has an AI powered document comparison tool. Natural England is employing AI to revolutionise its mapping. HMRC is using AI in its live chat function – and for those wrestling with their tax forms but willing and able to use such a service, this will certainly help avoid the dreaded wait on the phone.

Later this week, a stone’s throw from the Treasury and HMRC, the House of Lords debated an Artificial Intelligence (Regulation) Bill proposed by the Conservative peer, Lord Holmes of Richmond. As with all new technologies, correctly crafting the initial regulation and legal framework for its deployment will be vital – and there is a cross-section of views in the Lords on the details, though much agreement on the need for careful legislation. The debate is often disproportionately focused on the perceived risks or threats posed by privately developed AI technologies. But the potential for the public sector to play its part in the innovation of AI and its use cases should be highlighted and the government can do more than simply act as a legislator and regulator. Insofar as this Bill proposes a structure for sandboxing new AI models within the supervision of the proposed AI authority, the public sector itself should be included as a potential sandboxer, to provide a testing environment.

Innovation in AI is not the preserve of private enterprise. Government and the public sector have a vital role to play, so as well as working more closely with the AI industry, the government must recruit, train and employ more of its own AI literate officials and technologists fully to unleash the potential for AI to revolutionise the provision of government services. Moreover, although the focus on safety and the avoidance of harm rightly remains central, it should not be allowed to hold back the potential for these technologies to help fix deep-rooted problems in our public sector.

Lord Holmes’s Bill may run out of time in this parliament. But effective AI regulation matters if the Treasury’s ambitious public sector productivity plan is to be realised. This ‘AI moment’ is a huge opportunity to transform the speed, volume and quality of public sector outputs, and finally get public sector productivity out of its slump, an issue that has dogged governments for years. The question is whether the government is brave enough to do it to deliver for its ‘service users’ – that is, you and me.


Lord Kempsell

Lord Kempsell was previously Political Director of the Conservative Party and Director of the Conservative Research Department. He has extensive experience advising leaders on strategy, campaigning, and communications at the highest level. Previously he was Special Adviser to the Prime Minister in the No10 Policy Unit where he delivered cross-government projects to improve the use of evidence and data in policymaking. Prior to working in politics, he was a journalist, and now writes freelance for titles including the Daily Mail and the Express. He sits in the House of Lords as a Conservative peer.

View All Posts