AI adoption is being enthusiastically promoted across the public sector, often driven by a desire to address fiscal challenges, improve efficiency, and transform service delivery.
In many ways this enthusiasm for AI solutions makes sense, given the challenging fiscal environment government is operating in. Public services face substantial budget pressures, and digital technologies are often presented as a way to “do more with less.”
However, UNISON is concerned that this approach may prioritise cost-cutting over genuine service improvement, and that AI solutions may be introduced to the public sector without sufficient assessment of their risks or evidence of their effectiveness.
While AI offers potential benefits for public services, its implementation must be carefully managed to protect workers, citizens, and service quality.
How AI is being deployed
For UNISON members working in public services, AI is not an abstract future possibility – it’s already changing how services are delivered and how work is organised.
Nearly half of public sector professionals report awareness of generative AI usage within their area of work, and over two-thirds of government bodies are piloting or planning AI implementation.
- Central government departments are increasingly using AI-powered tools for policy analysis, pattern recognition in large datasets, and citizen services
- Local authorities are adopting tools like Microsoft Copilot, using chatbots for initial citizen inquiries, and employing data analytics to predict service needs
- The NHS is developing AI diagnostic tools to improve diagnosis speed and accuracy, while using automation to streamline administrative processes and manage patient flow
Risks and challenges
While AI offers potential benefits, it also presents significant risks when deployed in public services.
Automated decision-making affecting people using services
When AI makes or assists with decisions that affect service users, particularly vulnerable individuals, any errors or biases can have serious consequences. For example, flawed algorithmic assessments of needs could lead to inappropriate care packages or resource allocation.
Inaccuracies and bias
AI systems can produce incorrect outputs, sometimes called “hallucinations.” They may also perpetuate biases present in their training data, potentially leading to discriminatory outcomes in service delivery.
Lack of transparency
Many AI systems operate as “black boxes,” making decisions in ways that are difficult to understand or challenge. This lack of transparency is particularly problematic in public services, where accountability is essential.
Data privacy concerns
AI systems require vast amounts of data to function effectively. In public services, this often means sensitive personal information about citizens, raising important questions about data security and privacy.
Impact on service quality and jobs
Without proper planning and worker involvement, AI implementation may negatively affect both service quality and working conditions. Workers may face increased job insecurity, intensified workloads, or a loss of autonomy and discretion.
Ethical and environmental implications
AI uses huge amounts of power and has significant environmental consequences, while also presenting a number of ethical problems such as copyright and intellectual property concerns. Its development is also associated with unfair and exploitative labour practices, often in the Global South.
