Where can I find out more?
If you require any further support in addressing ethical issues in your planned project or incorporating ethical principles, please contact the UK Statistics Authority Data Ethics team or visit our webpages for further information on the support services that we offer.
Thig guidance has focussed on some of the ethical issues relating to the use of machine learning in the context of research and statistics – there are a wealth of other sources available that relate to machine learning ethics, and machine learning more generally.
The United Nations Economic Commission for Europe (UNECE) – Machine Learning Group 2021:
The Machine Learning 2021 Group, led by the United Kingdom’s Data Science Campus in collaboration with UNECE is a collaborative working group, with a focus on modernising the production of statistics, specifically in relation to machine learning. The group is made up of experts from statistical institutions around the world, who come together to share their knowledge and research, in order to develop our understanding of machine learning.
You can find out more information at the following sources:
- UK Statistics Authority’s ethical principles and ethics self-assessment tool.
- The Cabinet Office, Central Digital and Data Office, and Office for Artificial Intelligence provide an Ethics, Transparency and Accountability Framework for Automated Decision-Making.
- The Alan Turing Institute provide research and guidance on ethical considerations of machine learning and artificial intelligence. Specifically, they have produced a guide for the responsible design and implementation of AI systems in the public sector.
- The Institute for Ethical AI & Machine Learning carries out research which considers the ethical and responsible development, deployment and operation of machine learning.
- Quality Assurance of Code for Analysis and Research.
- Central Digital and Data Office data ethics framework.
- The Open Data Institute provide a number of resources relating to the ethics of machine learning.
- Digital Catapult have created an applied methodology for machine learning ethics, detailing lessons learned from machine learning programmes.
- The European Commission have created a framework of ethics principles to use in relation to artificial intelligence, and a self-assessment checklist that guides users to put the seven principles into practice.
- The Australian Department of Industry, Science, Energy and Resources, have also created an artificial intelligence ethics framework for the public and private sectors.
- The Office for National Statistics’ Data Science Campus has some interesting examples of machine learning projects focused on the research and statistical space.
- The Government Statistical Service, Good Practice Team, have produced a user engagement strategy for statistics, which provides guidance on developing effective engagement activities within research and statistics.
- The Data Science Campus have a number of examples of machine learning projects that they have been working on, and provide information, blog pieces and training, relating to machine learning.
- The Office for Statistics Regulation have produced guidance on ensuring statistical models command public confidence, in response to the exam grading issues experienced in 2020. More recently, they have also published guidance on good practice in designing, developing and using models.
- The Information Commissioner’s Office have produced guidance on how to audit artificial intelligence focusing on best practices for data protection compliance.
- The Government’s Office for Artificial Intelligence has produced a number of documents relating to machine learning. The Centre for Data Ethics and Innovation have produced a number of blogs and articles relating to machine learning, including those on trustworthy data governance and addressing misinformation on social media platforms.
- The Alan Turing Institute have produced and collated a number of resources which relate to machine learning and artificial intelligence more generally. This includes training resources, a Data Ethics Group, and guidance.
- The Ada Lovelace Institute have a number of projects and resources on their website, and have also set up “Just AI”– an independent network of researchers and practitioners committed to understanding the social and ethical value of data and AI.
- The Royal Statistical Society have begun a new “Data Science Section Initiative” which allows researchers to share and discuss ethical challenges encountered in the data science profession.
- The Open Data Institute have created some interesting guidance on the anonymisation of data in times of crisis, amongst other resources. The Atlas of AI discusses some of the important issues and considerations of artificial intelligence, including the use of natural resources, privacy, equality, and freedom.
- This guidance briefly mentions the environmental impact of machine learning models. Cornell University have created an emissions tool, and associated guidance to approximate these emissions.
- There are a number of really interesting pieces on the explainability of artificial intelligence, and how to make black box algorithms explainable.
- The Massachusetts Institute of Technology discusses the environmental impact of training models in their blog.
- Further academic articles relating to the carbon impact of machine learning can also be found here, and The Alan Turing Institute also provide a number of relevant resources.
- There are some very interesting articles which address the uncertainty of machine learning predictions.