Ethics watchdog to examine use of AI in public services
Committee on Standards in Public Life to conduct review of moral and technical issues facing use of new technology
The government’s public standards watchdog has announced a review of the use of artificial intelligence in public services.
The inquiry will look at how the increasing uptake of “technologically assisted decision-making” in the public sector is affecting service delivery, and whether existing regulations are sufficient to ensure standards are being upheld, the Committee on Standards in Public Life said this week.
It will examine how autonomous systems are being used to make decisions or solve problems that would otherwise require human intervention. This includes the use of algorithms in decision-making, which is the subject of a separate investigation announced last week by the Centre for Data Ethics and Innovation and the Cabinet Office’s Race Disparity Unit.
The committee’s review will examine public bodies’ adoption of new technologies, with a focus on the application of the standards embodied in the “seven principles of public life” – commonly known as the Nolan principles – which apply to all public sector employees. The Nolan principles are: honesty; integrity; objectivity; openness; leadership; selflessness; and accountability.
The non-departmental public body – the purpose of which is to advise serving prime ministers on ethical standard throughout public life – will also consider whether it is appropriate or in the public interest to use AI-assisted decision-making in particular contexts.
- Government data ethics body to focus on bias and micro-targeting
- Government places £1bn bet on AI
- Public ethics body to examine Home Office data use
Announcing the review, committee chair Lord Jonathan Evans said the Nolan principles had “stood the test of time” and had been adopted throughout the civil service and public bodies such as NHS trusts and local councils. “The increasing development and use of data and data-enabled technologies in our public services can potentially bring huge advantages in terms of pace and scale of service delivery," he said.
"But there are some major ethical and practical challenges about what this means for accountability, objectivity and the other Nolan principles."
Based on its inquiry, the committee will produce a set of recommendations to ensure public bodies maintain high standards in AI-assisted services and decision-making. It may put forward guidance on best practice or suggest changes to regulations.
The committee is seeking written submissions from individuals and organisations involved in developing policy, systems or safeguards on the use of AI. It is expected to report on its findings early next year.
This review will be among the first major pieces of work the committee has carried out since former MI5 boss Lord Jonathan Evans took up the role of chair last year.
During a pre-appointment hearing for the role in October, Evans said he would not shy away from the criticising the government should his position call for it.
He also said he wanted civil servants to be given the authority to assess risks and make informed judgements on contentious issues, rather than unthinkingly following established processes.
“If you pile bureaucratic process on bureaucratic process then the whole thing becomes a rather deadening compliance exercise,” he told the Public Administration and Constitutional Affairs Committee. “Where you are in uncharted territory or where there is an ambiguity, I feel it is much better if people feel a sense of responsibility to do the right thing."
Airline facing massive penalty after incident in which regulator claims the personal data of 500,000 customers was compromised
Force expresses disappointment in ‘negative and unbalanced’ tone of report it commissioned
Regulator publishes Openness by Design three-year plan
Ability to compel telecoms firms to gather data was introduced in Investigatory Powers Act
Sharon Hobson of Riverbed explains why the key to justifying an investment in cloud technology is visibility of network performance