A week after the UK government's announcement Big plan to make big investments in AI; It has revealed details of how it will shape up in the public sector. On the agenda: AI assistants to speed up public services; data sharing agreements across shelters; and a new AI toolkit named “Humphrey” after a character on an old UK TV political sitcom to speed up the work of civil servants.
Science The plans will be officially unveiled on Tuesday, led by the Department for Innovation and Technology (DSIT), along with two other departments – Work and Pensions and Health/Social Care.
If navigation The UK Government's AI site to check the progress of certain projects; You will find that most of the efforts so far have been limited experiments or in the early stages of testing. Others are even more grounded. For example, A plan to bring more AI services to the customer-facing side of the NHS is “just one step.Charter“I follow the concept.
Some include links to Github repositories to check out some of the work to date. How many total people are working on these projects? It is also unclear which third-party tools (such as LLMs) are being used. (We asked these questions and will update as we learn more.)
At their heart, projects are about performance. DSIT said the government currently spends around £23 billion annually on technology and is looking to repurpose that money in more innovative ways.
“Slow technology is a long-term and costly impediment to our public services… the headaches left behind after the pause. Not to mention stress and forced travel to replenish. A pattern,” DSIT's Secretary of State Peter Kyle said in a statement. “My department will put AI to work, use technology to help us tolerate the nonsensical approach being used by the public sector, and use it to work together to help the people it serves.”
Plans include a new team within DSIT to lead similar projects. DOGE In the US, however, it is run by government people rather than technocrats.
DSIT is initially focusing on three areas.
1. Work of government employees. Named after Humphrey. angry Clever assistant Starring the late Nigel Hawthorne, “Yes, Prime Minister” is a suite of apps aimed at reducing the daily work hours of employees, especially those with large amounts of data they need to read. And as part of their business.
The “consultant” is designed to read and summarize “thousands” of responses to consultations in a matter of hours (responses, which can be lengthy and voluminous, are a key part of how the government considers feedback from stakeholders and the public). “Parlex” allows users to query and read conversations in parliament related to bills or other policy documents they are working on. “Minutes” is a secure transcription service to take notes from their multi-hour meetings. “Redbox” helps prepare briefings and policy documents. And “Lex” allows you to consult relevant legal information.
2. Another type of efficiency drive has to do with speeding up public services. The idea here is to address the legacy bureaucracy of which there is so much in the UK. The 100,000 calls the tax authorities receive every day, or the need to appear in person to register a death, or (extraordinarily) place ads in local papers as part of the process of obtaining a license to drive a truck.
DSIT's thinking is that streamlining processes like this through AI-fueled automation could save £45 billion annually. (This estimate is unclear before or before deducting the cost of building and operating AI services.)
3. The last part will be more collaboration between departments to help share data to speed up how services are accessed and how they work.
The various projects are a signal that the government is referring to businesses for their new AI push. But it also raises many questions.
For example, In the case of data sharing; For now, DSIT says the operational idea here will be “a common approach to information sharing.” Central government departments such as HMRC (Hime and Customs) and the Department of Business and Trade can share data with each other and with local councils in fraud investigations or to better understand how businesses are doing and how they can help. But what happens to data protection for individuals when you share data without intending it to happen?
Another possible question concerns Humphrey: right now, DSIT says some of the early apps are still in the testing phase, but the big question is how far the government will go to trust some of AI's conclusions.
There will also be human challenges. As a former civil servant (now working at an AI company), NotesPast efforts to create programs that cut across departments have not always worked. cooperation, Money and power are ultimately the forces that will make or break any of these plans.