The Government’s AI Accelerator Programme is training public sector data scientists to build machine learning tools, which is excellent news for public services and devastating for the office keyboard that had spent years believing it was the department’s most advanced system.
Announced on 28 March 2025, the 12 week programme brings up to 25 participants from areas including justice, health and transport together to work on live public service challenges. The keyboard has responded by requesting a learning plan, a mentor and a lanyard identifying it as a digital enabler.
The programme aims to improve services, reduce backlogs and find efficiencies. It will also expose civil servants to the dangerous idea that a process can be changed without first creating a steering group to discuss whether changing it might affect the process.
At the College of Cautious Algorithms, training steward Rita Boolean welcomed the initiative. “Machine learning needs capable people,” she said. “We have therefore given the capable people more diagrams and asked the diagrams to explain themselves.”
Responsible use of AI is central to the course, with participants considering transparency, accountability and bias. A nearby spreadsheet has taken this personally and is now asking every cell to declare whether it has benefited from structural privilege elsewhere in the workbook.
The data scientists are expected to develop tools that can eventually be deployed in their departments. This has alarmed the office printer, which has submitted a proposal to automate the message paper jam after checking that there is no paper jam.
Director of Useful Computation Rollover Smith said the programme would help officials understand how data could improve public services. The mouse has already volunteered for a technical secondment, although it remains unclear whether it can be trusted with a second button.
The keyboard completed its first module on machine learning ethics this morning. It now types synergy only when nobody is watching, delivering the first productivity saving by ensuring that at least one machine in government has learned when to stop talking.
Real story: Department for Science, Innovation and Technology.
