The answer to life, the universe and everything was, according to Deep Thought, the supercomputer in Douglas Adams’ Hitchhiker’s Guide to the Galaxy, 42. For government, the answers to everything from making the NHS more effective and back-office procedures more efficient to procurement and mitigating climate change are proving more complex. Artificial intelligence, however, is being increasingly promoted as a game changer and was the subject of a day-long event attended by top-tier civil servants and hosted by IBM in partnership with Civil Service World.
The conference asked a simple question: Rethinking Public Services: Can AI Deliver a Better Future? The future is already here, if in a rather embryonic form: according to Ruth Kelly, chief analyst for the National Audit Office, just 30% of 90 governmental departments and arm's-length bodies use AI for some form of digital application, though 70% are “actively planning for AI”.
The key for government is to recognise that AI will re-shape the way it works to be more responsive to the needs of citizens, said Giles Hartwright, associate partner, defence, at IBM Consulting. “Citizens increasingly expect digitalised services, so scalable and secure technologies will be the backbone of modern governance.” For now, many datasets are in dire need of modernisation – the Police National Computer, in use by all frontline police forces, Hartwright pointed out, dates to 1974.
Moving from rigid bureaucracy to a more dynamic structure would require “ongoing training and fostering a culture of innovation”, Hartwright added, noting the public sector would also need private and academic input. “We need agility and collaboration.”
Speaking of his own time in the civil service, Hartwright noted that “collaboration, historically has not been the greatest quality of government departments.” The skills base of the civil service also needs to be improved, he said, as just 4% of civil servants are digital professionals, compared to 8-12% in wider industry.
A new era in government
Despite the challenges, this was an exciting time to be in the civil service, David Knott, the UK’s chief technology officer declared. “We’re moving from an era of experimentation into an era of implementation,” he said, saying it was time to build on the foundational work of the Government Digital Service. This will mean, he suggested, moving from a time when the focus was on wide experimentation, to a more focused approach on the areas where AI could make the most difference.
“Lots of things are up to date in government departments but we have to make that normal across departments.” Earlier this year, a report from the Alan Turing Institute found that AI could help automate around 84% of repetitive transactions across 200 government services.
Government also needed to focus on addressing other fundamental barriers to AI adoption, such as legacy systems, data quality and encouraging a digital-first culture. Knott asked: “How do you change the culture? We need long-term investment and vision. There is a need to identify digital leaders and give them the mandate to implement change in departments.”
AI in practice
A fascinating example of how AI can make a meaningful difference came from Jan Spoenle, Presidential Judge at the Court of Appeals in Stuttgart in Germany. AI proved transformational in dealing with a deluge of legal claims concerning the “Dieselgate” car manufacturers’ scandal. Cases increased 10-fold in just five years, nearly every case went to appeal and submissions could stretch to 180 pages. “Unfortunately, we were not able to appoint 10 times as many judges to deal with everything,” he said. “We had to think outside the box.”
A human, manual process to identify the “data diamonds” – the key data required to uphold or dismiss a case, such as engine type or emissions standards – would have proved incredibly time-consuming. Instead, the judiciary worked with IBM to develop and deploy an AI system called OLGA to handle the repetitive tasks that did not require legal expertise, such as searching for key words. The system featured a dashboard of case files and filter tables such as car brands or legal firms involved. “OLGA was in effect an intelligent research assistant and easy to use. We told the judges if you can buy things on Amazon, you can use OLGA,” said Spoenle. Crucially, OLGA was not bespoke to Dieselgate. “We wanted to design an AI tool that could be applied to other issues – such as claims against online gambling companies.”
People power
One concern around AI can be the potential for job losses, but Spoenle was reassuring. “In reality, we actually had too few people originally, so by using OLGA, the existing overworked staff were freed up to work on more useful tasks where humans need to be in the loop.” He also noted that by removing the most repetitive parts of human work, systems like OLGA could help to address recruitment and retention challenges faced by public sectors across many countries.
This was also the experience of NatWest AI Product Manager Mark Worden, whose bank’s virtual assistant CORA has freed up staff to have longer, more meaningful chats with customers, as the technology’s language models pass customers onto a human being for complex or sensitive subjects, such as bereavements.
The benefits were clear to Blake Bower, director of Government and Regulatory Affairs, IBM UK: “People go into the civil service to make a difference, they don’t want to spend time doing the mundane legacy tasks, AI can be motivational, freeing them up.”
The skills gap
Not only will AI affect how civil servants do their jobs; for technology to bring about real change and drive long-term productivity, it must be part of a broader shift in how government and public services work. AI will re-shape what it is to be a public servant, with new skills and new roles, according to Kelly.
“AI by itself is no panacea,” she said. “Change is really difficult and challenging, you are not just replacing workload, you are redesigning the way in which you work, that is a big gap to cross.” To head off resistance to change, Hartwright argued that both technically and culturally, staff need to be involved early.
Viewing AI as a constantly rolling innovative approach, rather than as a project to be completed within a fixed timeframe, was helpful, suggested Bower: “As we develop policy the temptation is to think what legacy systems can support that – instead we need to think what support systems can we build alongside [policy] – it’s a different mindset.”
The NHS was given as a classic example of how AI can simultaneously help and expose fundamental structural shortcomings. Stuart Hoddinott, senior researcher at the IfG, cited a doctor who saves 20 minutes a day by getting AI to transcribe his notes, but then wastes an hour walking around the hospital looking for a printer that works. Data is also very fragmented, he added, with most trusts not fully digital and using systems that do not talk to each other.
All the speakers agreed, therefore, that it was important not to be dazzled by AI but to focus first on the practical tasks it could help with. For civil servants, much of the impact of AI will come first in relation to operational work rather than interacting with the public, said Kelly, and be applied to common applications and functionality such as facial recognition and cross-checking documents.
The day had begun with Hartwright presenting research from IBM’s Institute of Business Value which explored the characteristics of governments who are best able to seize the possibilities offered by new technologies. Over the course of the day discussion confirmed the importance of these characeristics: investing strategically in data and tackling legacy systems, empowering people to embrace change, and building strong partnerships with external partners. “Transformation in government is rarely achieved in isolation,” Hartwright said. “Trusted partnerships between the public and private sectors are essential to driving change... together we can create a public sector that not only meets today’s challenges but also leads the way in shaping a better future for everyone.”
The event was sponsored by IBM, who also sponsored this article and were given advance sight of it before publication.