Analysis

The year 2023 was frequently heralded as the year of AI. Given the remarkable advancements we saw over those 12 months, especially around generative AI, it’s not an unfair label. With 100 million users within just two months, innovations like ChatGPT were adopted at astonishing speed by people around the world.

Of course, AI is nothing new. In 1966, one of the very first AI chatbots – ELIZA – was developed by Joseph Weizenbaum. Prefiguring ChatGPT by nearly 60 years, it was understandably limited. Through the 1970s and 1980s, the field of AI saw breakthroughs, counterbalanced by periods of prolonged stagnation.

During the 1990s, progress was more sustained. The success of IBM’s Deep Blue supercomputer in 1997, defeating reigning world chess champion Garry Kasparov across six matches, was one notable moment that caught the public’s attention. Later victories, from IBM Watson in the game show ‘Jeopardy!’ to Alpha Go’s 2017 triumph over the world number one in ‘Go’, only added to the hype.

But the current wave of AI seems genuinely different. When we published our first Digital Leaders Study in 2021, there were no mass-adopted generative AI products. By the time we released our 2023 follow-up report, GPT-3.5 had been accessible for several months. As we publish this paper, we’ve seen two new models from OpenAI (GPT-4 and GPT-4o), demonstrations of remarkable text-to-video capabilities (through software like Sora), the full release of Google’s Gemini, and ever growing use among the public.

For governments and private companies the world over, it’s become ever clearer that AI is the future. The UK – both the state and its commercial enterprises – are now in a truly global race to make the most of AI.

But for all the (often justified) hype around generative AI, the defining feature of this new wave of tech is that it’s been highly disaggregated: individual citizens have led the charge in adopting
these tools. It’s not uncommon for someone’s personal use of ChatGPT (or similar tools) to far outstrip what’s being used in their workplace. And when it comes to larger institutions, implementation has been relatively limited.

For government, this is partly understandable. Unlike with the digital transformation we saw in Whitehall – when the state was playing catch-up with the private sector – the latest AI innovations aren’t yet widely adopted across many large businesses either. There’s much more limited best practice to copy from and risks are therefore greater. With fewer common standards or business change models to work from, there is some rational caution here, especially given the government’s implicit duty of care. All AI systems that are citizen-facing must be trustworthy and reliable, but must be even more so when people are interacting with the state and public services.

And yet, despite these valid concerns, government has to scale up AI and do so quickly – the opportunities are simply too vast to pursue a path of delay. Moving fast is essential. But this is set to be a much more difficult transition than with digital.

The UK and Singapore

Under the previous government, the UK began to develop a stronger AI leadership role. The creation of the Department of Science, Innovation and Technology (DSIT) in February 2023 was explicitly framed as “positioning the UK at the forefront of global scientific and technological advancement”. As the new department tasked with ‘policy for AI’, it subsequently promoted a new white paper, offering a “pro-innovation” model of AI regulation that contrasted with the
EU’s more restrictive approach.

The strongest leadership has been around AI safety. In November 2023, the UK hosted the first AI Safety Summit at Bletchley Park, attracting global leaders – from the president of the European Commission and the vice-president of the United States to OpenAI CEO Sam
Altman – and securing a multilateral agreement to test major AI models. The launch of the AI Safety Institute in February 2024 established the UK’s capability to carry out that work.

Maximising the opportunities of AI is now a point of policy consensus in the UK. Just as the previous prime minister Rishi Sunak said that AI would “bring a transformation as far-reaching as the Industrial Revolution”, the new government’s manifesto has committed to harnessing AI “to transform the speed and accuracy of diagnostic services” in the National Health Service. And global AI rankings show that the UK is building from a solid base. Oxford Insights places the UK
third in its international AI Readiness Index; Tortoise Media has us at fourth in its Global AI Index.

And yet, there remains a central tension in our approach – this is “the accelerator and the brake” metaphor explained earlier on. Though there’s much excitement among ministers and officials about AI, the bulk of the practical activity so far has been on that brake: to regulate, govern, set standards and create safeguards. This is all necessary and valuable, but without practical adoption of AI in government – the accelerator – we risk falling behind, when others are racing ahead.

There’s a useful contrast here with Singapore. The latter is ranked third globally in the Tortoise AI Index and second by Oxford Insights – in each case, one place above the UK. While our government published a single National AI Strategy back in 2021 (a document which makes no reference to generative AI), Singapore has already published two – with the most recent released last year.

There’s more evidence of a disparity here. In 2020, a collection of Singaporean agencies published a guide to redesigning jobs around AI – more than a vision alone, it’s closer to the design and plan components of the 7 Lenses. Crucially, Singapore has also gone further than the UK in implementing AI across public services. This is where the UK must get to: AI implementation, not just ideation.

Whatever the critics may say, the new UK government has taken power with an ambitious set of goals. Labour has set five overarching national missions – from securing the highest sustained growth in the G7 and delivering zero-carbon electricity by 2030, to reducing the number of lives lost to the most deadly diseases. Achieving this while fixing a catalogue of broken public services won’t be easy. But with a high tax burden and limited scope for more spending, operating differently is key.

AI must form a core part of this work. A frugal innovation approach – in which services are transformed in as cost-conscious a way as possible – will be essential.

The 10 interviews we conducted with digital leaders offer useful guidance for new prime minister Sir Keir Starmer and his team. They provide invaluable insights into how the UK can move closer to Singapore’s approach: from talking about AI, to actually putting it in place.