Jessica Elgot and Robert Booth 

AI tool can give ministers ‘vibe check’ on whether MPs will like policies

Parlex is one of several artificial intelligence systems being developed within the government
  
  

Keir Starmer in front of rows of MPs at PMQs
AI tools are being developed to predict which topics might cause difficulty with backbenchers. Photograph: House of Commons/AFP/Getty Images

A new artificial intelligence tool can warn ministers whether policies are likely to be very unpopular with their party’s MPs, using a search described as a “parliamentary vibe check”.

Parlex is one of a suite of AI tools being developed for ministers and civil servants which allows them to predict which topics might cause them difficulty with their own backbenchers – and pinpoint specific MPs who feel passionately about a given subject.

A summary of a policy – such as a 20mph speed limit – can be given to the tool which then predicts how MPs are likely to react, according to their previous contributions in parliament. A demonstration video on the government’s website shows how Tory MPs have historically opposed the change and the Labour MPs in favour of traffic calming measures.

The Parlex tool’s description says it “allows policy teams to understand the political climate and anticipate potential challenges or support for a policy before it is formally proposed and to build a parliamentary handling strategy”. Parlex, which is at the early stages of development, describes this as a “vibe check”.

The tool, first reported by the Times, shows that MPs including Iain Duncan Smith and the former MP Tobias Ellwood were against 20mph speed limits and that the Labour MP Kerry McCarthy had spoken in favour of traffic calming measures.

A government source said the tool was most likely to be of use to civil servants rather than ministers, who they would hope to be politically savvy enough to be aware of the views of well-known MPs.

The prime minister, Keir Starmer, launched an AI plan last week that ministers said would mean the technology would be “mainlined into the veins” of the nation, with a multibillion-pound investment in the UK’s computing capacity.

The government plan features a potentially controversial scheme to unlock public data to help fuel the growth of AI businesses. This includes anonymised NHS data, which will be available for “researchers and innovators” to train their AI models. The government says there would be “strong privacy-preserving safeguards” and the data would never be owned by private companies.

Ministers believe AI can help tackle Britain’s anaemic economic growth and deliver, according to their own forecasts, an economic boost rising to up to £470bn over the next decade.

Parlex is one of a number of new AI tools being developed in government. The Incubator for Artificial Intelligence inside No 10 has now moved to the Department for Science, Innovation and Technology (DSIT). The unit said at the end of last year there were 22 projects in incubation and 11 in the development pipeline.

Another tool in development which is more widespread is Redbox. It means civil servants can automate submissions to ministers by analysing government papers, including ones that are marked “sensitive”.

The tool is said to transform what originally took hours of manual work and will be given to all civil servants in the Cabinet Office and DSIT before an intended rollout to every civil servant.

Another programme, Consult, is being developed to save approximately £80m annually by automating and improving the consultation process, intended to allow civil servants to more effectively analyse and act on the views of the public.

The Department for Work and Pensions has been among the AI leaders in the government, including in its use of “white mail” which analyses handwritten letters from vulnerable people from the about 22,000 letters the department receives each day, meaning those most in need can be more quickly directed to the relevant person.

But there have also been major problems, including 100,000 people wrongly flagged for potential housing benefit fraud after the performance of a government algorithm fell far short of expectations.

 

Leave a Comment

Required fields are marked *

*

*