Guidance on using AI and algorithmic management in the Welsh public sector
This guidance has been developed and agreed by trade unions, public sector leaders and the Welsh Government through the Workforce Partnership Council. Its full title is Managing Technology That Manages People: A Social Partnership Approach to Algorithmic Management Systems in the Welsh Public Sector. It was approved by representatives of the NHS Confederation and the Welsh Local Government Association. All unions and public bodies have a mandate to adapt and adopt this guidance in their workplace.
About AI and algorithmic management systems
Artificial intelligence (AI) systems are being used in the workplace to automate tasks and to manage workers. The term ‘algorithmic management systems’ refers to AI and similar technologies which are used to manage staff.
What are the risks of AI and algorithmic management in the workplace?
Concerns regarding the right to respect for private life and data protection and the right to non-discrimination have been highlighted.
The cumulative effects of algorithmic management can damage the health, safety and well-being of workers: expectations and the pace of work intensify and the experience of being monitored closely has a negative impact on mental health.
We should also be mindful of the risk of a decline in the place of human judgment, human expertise and human connection in our working relationships.
Guiding principles
The following set of general principles that should be followed by unions and management throughout negotiation, decision-making, deployment and ongoing evaluation of any AI and algorithmic management systems.
- Social partnership
Unions to be involved at every stage of adopting, implementing and reviewing AI systems. This is in line with the Social Partnership and Public Procurement Act 2023 and long-established ways of working in Wales.
- Human oversight and human interaction
There must be human oversight of all strategic decisions taken about the use of AI and algorithmic management systems (a “human in command”) and clear lines of responsibility to management drawn for any decision taken. Human interaction should also be required in the day-to-day running and decision-making of a system (a “human in the loop”).
- Adherence to the Welsh Government’s definition of Fair Work, especially in relation to equality
According to the Welsh Government, Fair Work is the presence of observable conditions at work which means workers are fairly rewarded, heard and represented, secure and able to progress in a healthy, inclusive working environment where rights are respected.
- A commitment to training regarding AI algorithmic management systems across the workforce and at all levels
Managers and trade union representatives should be trained on how these systems work, the risks associated with their use and how to effectively monitor them. Furthermore, all staff should be trained on how to use and safely interact with these systems.
- A commitment to the protection of jobs, the creation of jobs, and investment in the workforce
AI systems should not be introduced with the aim of cutting numbers of staff. Where these systems improve efficiency and lead to tasks taking less time or requiring less human involvement, then staff should be trained on new tasks and jobs.
Guidance
Management should follow this guidance and include trade unions at every step.
Before Using AI and algorithmic management systems
- Check Existing Systems
- Before starting, check what AI and algorithmic systems are already being used.
- Use a standard way to record information about these systems, like how they work, why they are used, and who is responsible for them.
- Keep this information updated and share it with everyone in the organisation.
- Decide to Build or Buy
- Before deciding to create or buy a new system, check if the organisation is ready.
- Follow guidelines for buying or building AI systems and ensure they are secure.
- Ask important questions like whether the technology is needed and if the organisation is ready for the change.
During Development or Purchase
- Consult and Assess
- Work together with different groups to ensure the system respects important principles.
- Use a standard to discuss and record information about the new system.
- Prepare assessments to identify and address risks, like data protection and equality.
- Consider Risks
- Think about laws and rights, like health and safety, human rights, and data protection.
- Consider broader concerns like the impact on workers and the need for human judgment.
- Plan for the end of contracts with suppliers and ensure operations can continue smoothly.
- Involve trade unions
- Management must ensure that trade unions are involved in the process to provide feedback and help communicate changes.
- Management must give consideration to providing the training and support trade unions need to participate effectively in consultation on AI adoption.
After Implementation
- Monitor the System
- Regularly check if the system is working as intended and if it is affecting work conditions.
- Address any new risks or issues that arise.
- Ensure transparency and allow workers to raise concerns without fear.
- Human Oversight
- A senior manager should be responsible for the system and its decisions.
- Ensure decisions with significant effects involve human judgment.
- Protect Individual Rights
- Ensure individuals' rights are protected, like the right to access their data and request reviews of automated decisions.
- Agree on additional rights and safeguards for individuals.
- Share Information
- Share information about the system and any changes with all relevant parties.
- Consult with unions and representatives when making updates.