The Alan Turing Institute is teetering on the brink of collapse. In July, Technology Secretary Peter Kyle wrote a letter instructing the UK’s national AI institute to shutter the vast majority of its research programmes and refocus on defence and national security. This followed extensive criticism from researchers and former staff about its lack of impact on the UK’s AI ecosystem and failure to prepare the Government for major technological developments.

In the last week, Turing staffers have orchestrated a fresh round of media attacks and a Charity Commission complaint, alleging a “toxic internal culture”. This criticism comes from staffers aggrieved that their projects have been cut amid the institute’s funding woes, placing them at risk of redundancy. These projects, covering areas such as online safety and health inequality, are examples of work Kyle believes the institute should not be doing. Appeasing this faction to restore calm would move the Turing further from the Government’s preferred goals. Meanwhile, the institute risks losing disillusioned staff who specialise in defence and national security.

Rooted in UK universities, which have long been absent from pioneering AI research, the Turing was slow to recognise the importance of large language models, and often seemed to get lost in highly politicised ethics debates. Its defence work, ringfenced from the rest of the institute, drew on non-academics and partnered directly with the national security community. This team is more widely respected by those who work in AI than the rest of the institute.

Despite evidence of AI’s growing importance on the battlefield, the UK remains a laggard. A Select Committee report published earlier this year warned of “a gap between rhetoric and reality” in the Ministry of Defence’s adoption of AI, with “few examples of Defence AI applications making it beyond small-scale experimentation and actually being adopted by the Armed Forces”.

There is a case for the UK breaking its reliance on overseas labs or contractors and conducting its own R&D around autonomy for drones, electronic warfare, or sensors. The MoD might announce new innovation units, but its delivery record remains thin. Moving this work out of the bureaucratic Government machine and into a well-funded defence technology institute could help.

Yet the chances of this happening are slim, and receding further. Despite its status as a “national” AI institute, ministers have no direct power over the Turing. They do not pick its CEO or chair, play no role in setting its strategy, and have no say in its governance.

Power largely lies with the Engineering and Physical Sciences Research Council (EPSRC), the Turing’s primary funder. Under the Higher Education Act 2017, once a government has signed off on the strategy set by UK Research and Innovation (which oversees research councils), it cannot interfere in funding decisions. The EPSRC has expressed concern about the governance and effectiveness of the Turing, but its leadership is determined to avoid even the appearance that it is responsive to political pressure. Universities which also provide substantial funding are unlikely to support shutting Turing’s academic programmes, however dubious, in favour of national security work, with academics already decrying the Kyle letter as an attack on their freedom.

The Turing’s leadership has gambled that it can play the Government and the EPSRC off against each other until media and political interest subsides. The CEO and Chair have both told staff that they have no plans to alter the institute’s focus.

UK Research and Innovation’s present strategy runs until 2027 so, short of more drastic intervention, the Government can only hope that the EPSRC tires of the public feuding and acts unilaterally. The risk remains, however, that the Turing disintegrates before this happens, a possibility only made more likely by the complaints made by whistleblowers in recent days.

A secretary of state sending a letter pleading with a publicly-funded body to change its focus and said public body deciding to ignore it is a good metaphor for British state capacity, but Kyle is not the architect of this impasse. The Turing is a reflection of a Coalition-era aversion to serious thinking about institutions. Instead of investing in critical capabilities, the 2010-15 government accelerated the trend of outsourcing science and innovation policy to a collection of universities, research councils, and other vested interests under a haze of buzzwords about decentralisation. A well-functioning national institute, ultimately, requires national seriousness.

---

Alex Chalmers is an Editor at Works in Progress and the author of Chalmermagne, a Substack focused on policy, finance, and technology.

chalmermagne

---