In this biweekly bulletin, Roula Khalaf, the managing director of the FT, handpicks her preferred articles.
To streamline the UK legal service and enhance public sector efficiency, there is a proposal to establish an artificial intelligence “hit squad” unit at the heart of Whitehall.
Deputy Prime Minister Oliver Dowden plans to allocate an annual budget of approximately £5 million to assemble a team of 30 “highly skilled, essentially capable” AI and data engineering experts to kickstart the modernization of public services.
During a session on Monday discussing an AI training initiative for legal professionals in London, he lauded the technology as the “closest thing you have to a magic bullet in terms of enhancing performance for the payer.”
Dowden elaborated on the unit’s objectives, emphasizing its role in expediting the data collection process from diverse agencies, filtering out irrelevant information, and facilitating swift decision-making by officials.
He articulated the unit’s mission to form a task force—a specialized group that would offer advanced expertise to explore innovative solutions for various projects.
The unit’s primary focuses, according to Dowden, will include combating security fraud, streamlining asylum application procedures, and enhancing interactions between the general public and the NHS.
Given the documented instances of algorithms exacerbating racial biases, entrusting AI with decisions related to welfare fraud and asylum requests could spark controversy.
Amnesty International highlighted a case in the Netherlands where a self-learning system implicated in a caregiver dispute was found to exhibit racial profiling, leading to government repercussions in 2021.
Dowden underscored that the technology would serve as a “co-pilot” within the UK public sector, complementing human decision-making processes.
Laura Gilbert, the chief scientist and data research chair at Downing Street, stressed the importance of maintaining human oversight in AI applications, affirming that “the people in the loop are crucial.”
Dowden outlined plans to leverage AI for contact management and correspondence, citing its potential to “significantly enhance efficiencies,” thereby enabling further downsizing of the legal service workforce, which is already projected to shrink by 66,000 positions.
He contended that AI could yield greater outcomes with fewer resources, positioning it as a substantial force in reducing Whitehall’s workforce, which expanded significantly due to Brexit and the pandemic.
While acknowledging the trade-offs inherent in government decisions, Dowden asserted that AI could bolster outputs, improve services, and lower costs simultaneously.
Despite skepticism from economists regarding AI’s capacity to boost public sector productivity, Dowden advocated for prioritizing fundamental IT system enhancements, increased investment, and management capacity building as alternative solutions.
Gilbert suggested that AI’s automation of administrative tasks could lead to a more content workforce in Whitehall, citing instances where officers spend an entire evening monthly on administrative duties.
Rather than pursuing a broad, all-encompassing cross-department AI strategy likened to “boiling the ocean,” Dowden proposed focusing on initiating approximately five major projects and ten minor projects to maximize impact.
Gilbert disclosed that the government presently utilizes around 20 open-source models and intends to incorporate AI systems from leading private entities like OpenAI and Anthropic as needed.
She hinted at exploring diverse AI applications beyond extensive language models to further enhance governmental operations.
