Can AI Help With Equality Impact Assessments? What Officers Need to Know

AI tools are entering local government faster than most equality teams anticipated. From chatbots handling resident enquiries to large language models drafting policy documents, the technology is already in use — often without clear governance frameworks for how it should or shouldn't be used in sensitive analytical work.

For equality officers, the question is not whether AI will become part of the EIA landscape. It already is. The question is how to engage with it in ways that maintain — or ideally improve — the quality and accountability of equality analysis.

What AI can genuinely do well

Current AI tools, particularly large language models trained on extensive bodies of research, can perform several tasks relevant to EIA work with genuine competence:

Synthesising research evidence. An AI system with access to a curated knowledge base of research from organisations like Age UK, Mencap, the Fawcett Society, and the Runnymede Trust can identify and summarise relevant findings far faster than a human researcher working through the same material manually.

Structuring analysis across all nine characteristics. One of the most time-consuming aspects of EIA work is ensuring comprehensive coverage. AI can generate systematic first-draft analysis across all nine protected characteristics, flagging areas where evidence suggests significant impact and areas where evidence is thin or absent.

Identifying likely impacts from a proposal description. Given a description of a council proposal, AI tools can draw on research evidence to identify the populations likely to be affected and the mechanisms by which impact may occur — providing a useful starting framework for officer review.

What AI cannot do — and what remains the officer's job

Understanding the limits of AI in this context is as important as understanding its capabilities.

AI cannot replace local knowledge. An AI system does not know that the ward most affected by a proposed leisure centre closure has an unusually high concentration of residents with physical disabilities, or that a specific community group has been raising concerns about access for the past year. Local context, community relationships, and knowledge built through lived professional experience cannot be generated by a language model.

AI cannot consult with affected communities. The PSED's expectation of genuine engagement cannot be discharged by AI. Officers, elected members, and community partners must do that work.

AI outputs require professional review. AI can produce a well-structured draft that cites relevant research. It cannot guarantee that the analysis is accurate, proportionate, or appropriate to the specific proposal. Officer review, contextualisation, and professional judgment remain essential.

Questions to ask when evaluating AI tools for EIA work

Not all AI tools are equal in quality or appropriateness for equality work. When evaluating any AI tool for use in EIA processes, equality officers should ask:

• What is the evidence base the system draws on? Is it curated, current, and from credible sources — or is it drawing on the open web, which contains significant amounts of inaccurate and low-quality content?
• Does the system cite its sources? An AI that produces analysis without identifying where the evidence comes from cannot be used to produce a defensible EIA.
• How is our council's data handled? Proposal descriptions may contain sensitive or commercially sensitive information. Understanding the data processing arrangements is essential.
• What governance framework applies to AI use in our council? Many councils are still developing their AI governance policies. Using AI tools for sensitive analytical work without a clear framework is a risk.

The likely trajectory

AI tools purpose-built for equality work — with curated evidence bases, transparent citation, and outputs designed for officer review rather than direct publication — represent a genuinely useful development for a function that has been chronically under-resourced.

Used well, they can free equality officers from hours of research and first-draft writing, allowing specialist expertise to be focused where it adds most value: reviewing outputs, adding local context, engaging with communities, and ensuring that equality analysis genuinely informs decisions.

The technology will not solve the structural resourcing challenges facing equality functions in local government. But for officers trying to maintain quality under significant time pressure, it is a tool worth understanding and engaging with carefully.