By Sam Peters, Chief Product Officer, ISMS.online
2025 is set to be the year in which the government and public sector entities more actively embrace AI within their operations.
Back in October 2024, Labour MP Nick Smith revealed that the House Authorities, alongside the House of Lords administration and Parliamentary Digital Service (PDS), have been actively exploring the potential for AI and other new technologies to support the work of Members.
“PDS is undertaking a cost-benefit analysis of Microsoft’s Co-Pilot AI tool, which includes AI for mailboxes,” Smith stated, acknowledging the potential for such tools to support staff in their management of day-to-day administrative activities.
It’s a noteworthy headline. Indeed, the PDS considering the use of AI tools to assist MPs reflects the growing integration of AI into public services and the desire to increase efficiency and improve service delivery.
Indeed, back in January 2024, the head of the National Audit Office, Gareth Davies, gave a speech in parliament in which he outlined that greater productivity could release tens of billions for government priorities, noting the important role that technology could play to “transform service delivery, reduce costs and improve the user experience”.
With budgets under scrutiny, the government must be seen to be practicing what it preaches, working to ensure every penny goes further. Clearly, this latest move from the PDS appears to be a step in that direction.
Effective governance is just as important as AI itself
Like many other sectors, AI has the potential to significantly modernise and improve efficiency in governance within parliamentary functions.
Indeed, AI could be used to deliver a variety of benefits that support the ambitions laid out by Gareth Davies, from enhancing productivity and increasing efficiency to optimising decision-making and improving service delivery. Yet despite the benefits, there are several key considerations that public sector entities must keep in mind when developing and deploying AI systems.
Data privacy, for example, must be made a key priority, with individuals increasingly seeking assurances that their sensitive and personal information won’t be misused by AI systems – a risk that Smith acknowledged during his speech.
“Before a pilot of Co-Pilot could happen within Parliament, there are important information rights protections to put in place to make sure that sensitive information is handled appropriately,” he stated, affirming that steps are being taken to put those protections in place.
In this sense, good governance will be just as important as the technologies themselves, with regulations now also emerging to drive the adoption of ethical AI development and deployment and safeguard sensitive information.
The European Union is leading this charge, with the EU AI Act being one of the first and comprehensive such regulations, focused on the protection of digital rights, fairness, and the elimination of algorithmic bias. However, such legislation isn’t without its critics.
Some circles argue that the EU AI Act is too strict, serving to dissuade investments into AI development due to excessive compliance demands. In the private sector, frictions have already begun to emerge, with both Apple and Meta having refused to sign the EU’s AI pact. Further, in June 2023, the former of these tech giants announced delays to the release of three new key AI features in Europe, citing “regulatory uncertainties”.
PDS: An example for others to follow?
Clearly the challenge for AI regulators is to strike the right balance between sustaining public safety without impeding the potential merits that these technologies can provide.
For this reason, in the case of the UK, the PDS’s exploration of AI tools is particularly interesting. Given that it is the responsibility of the government to strike this balance, parliament’s own approach to AI adoption will be closely watched, setting an example which others may follow.
Scrutiny or not, the PDS will have much to consider to get this right, with some of the key priorities likely to include:
- Ensuring that important government and public data is protected.
- Ensuring that it adheres to GDPR alongside other relevant legislation.
- Ensuring that AI tools are driven by transparent, auditable decision-making processes so that constituents can understand how AI is being used to serve them.
- Ensure that algorithmic biases are avoided in critical public-sector applications.
Data privacy and compliance are incredibly sensitive issues, yet ticking each of these vital boxes will not be easy. In effort to do so, it would be wise for the PDS to lean heavily on establish frameworks that have been specifically designed to guide organisations in achieving best AI practices.
Looking to key standards: ISO 42001
Enter ISO 42001 – the latest standard for AI management systems that aims to help ensure their responsible development, deployment and operation.
Placing significant emphasis on factors such as transparency, accountability, bias identification and mitigation, safety and privacy, its goal is to ensure that AI systems are built and implemented safely and ethically.
For the PDS, ISO 42001 could provide a roadmap that may be used to identify, evaluate and mitigate the risks associated with AI, ensuring that they can properly protect important government information and constituent data. Equally, it can provide a pathway through which the PDS can streamline its AI processes, identify and rectify vulnerabilities earlier, and reduce the potential financial and reputational costs associated with AI failures.
At ISMS.online, we recommend leaning heavily on ISO 42001 for good reason: it is likely to become a key benchmark for AI management systems in the future, and so leveraging it now will enable organisations and departments of all shapes and sizes to embrace best practices from the outset, rather than playing catch up or taking action only in the wake of a crisis.
By adhering to these compliance standards and addressing ethical considerations properly, parliament will be well placed to get this right and set a benchmark for responsible AI adoption that other government departments and organisations across the country can in turn follow.
Recent Comments