Question to the Department for Business, Energy and Industrial Strategy:

To ask the Secretary of State for Business, Energy and Industrial Strategy, pursuant to the Answer of 13 November 2020 to Question 112077, Artificial Intelligence, what progress has been made with each of the items listed.

A progress update where available (in italics) on each of the items in the answer I gave the Hon. Member on 13 November 2022 to Question 112077 can be found below.

BEIS are exploring AI and machine learning techniques internally to enable more efficient working. Projects are being

(i) Undertaken:

• A proof of concept for the use of virtual assistants to help staff find information regarding corporate policies, whereby the assistant will improve by learning from the enquires responses. An HR chatbot was developed as a proof of concept.
• Planning a proof of concept using Machine Learning for automatic labelling, setting up retention periods for past and future documents that form the official record. A small-scale proof of concept was undertaken in Spring 2021 using a "machine teaching" tool to automate content processing and data classification to help identify information of value within a digital heap.

(ii) Considered:

• The use of AI handling of inbound enquiries into the department to create draft responses and to triage requests to the correct teams. An enquiries service using an AI builder to automatically categorise emails was developed as a proof of concept.

BEIS Analysts use machine learning techniques, under the umbrella of artificial intelligence, where appropriate as part of analysis supporting policy development.

Machine Learning projects are being

(i) Undertaken:

• Identifying the location of industrial strengths. Report on this has been published
• Pilot for targeting communications about business support
• Categorising internal documents by subject. Currently exploring feasibility for implementation
• Project to understand the labour market through analysing job adverts. Methodology shared with OGDs and academics
• A pilot for organising internal processes. Process implemented

(ii) Considered:

• A pilot for predicting economic impacts using real time indicators
• Exploring automatic text generation a pilot exploring natural language processing approaches for extracting economic intelligence
• Planning to repeat a machine learning exercise on HMRC data to identify high growth potential businesses, to build on the successful ‘DECA pilot’ of 2019. This would underpin further operations in 2021, depending on the outcome of the SR process. DECA activities have pivoted to other government priorities.

BEIS policy teams are exploring the use of Artificial Intelligence. AI projects are being considered by the Better Regulation Executive who are looking to convert the stock of regulatory requirements placed on business into machine readable code and pilot hosting this as open source a metadata set on the ‘Open Regulation Platform’ (ORP), freely available on The National Archives GOV.UK platform. The project is currently in discovery phase to identify all data that government holds on regulatory obligations that could be relevant for this platform. This application is closely related to work that has already been undertaken as part of BEIS GovTech challenge to apply Artificial Intelligence (AI) to understand the cumulative impact of regulation. The Open Regulation Platform project is currently at the private beta stage.