Lessons from education unions building power on AI
Organised workers represent an essential source of countervailing power participate in, shape and contest technology at work. The challenge is to boost the movement’s capacity to meet the moment as AI technologies impact workplaces.
In partnership with nine unions, Connected by Data and the TUC worked across 7 months on an ‘action learning’ programme to test how to catalyse the movement’s capacity to understand, negotiate and advocate on AI from worker’s perspectives.
We sought to foster collaboration between education unions and other actors, and to gain clarity on key union issues on AI. Education was selected given its high political salience, contested policy environment and momentum already underway from education unions.
By the end of the programme:
the unions had produced a joint public statement on AI in education, establishing a common platform that fed directly into government consultations and external engagement.
Individual officers reported significant gains in knowledge and confidence.
Several unions restructured internally as a direct result, creating dedicated AI leads and breaking down siloes between policy and workplace-organising focused teams.
All nine ended the project with substantive work underway.
Along with articulating specific challenges for unions in the public sector, together we identified three considerations for building union capacity on AI.
First, programme design involves a trade-off between breadth and depth. For example a deep focus on collective bargaining produces momentum, while a wider spread of issues helps unions develop a more comprehensive approach but with less immediate impact.
Second, whereas a diversity of participating union roles can add a range of perspectives it can be tricky to translate discussion into shared goals and actions.
Third and most importantly, the programme affirmed that a context-driven approach to AI can help unions scope out the noise that surrounds the topic and focus work on the specific issues that AI presents for members.
Executive summary
A key leverage point to steer the technologies and governance of AI to be more purposeful, fair and effective is to build the countervailing power of workers and their unions to participate, shape and contest technology at work.
In partnership with nine unions, Connected by Data and the TUC worked across 7 months on an ‘action learning’ programme to test how to catalyse the movement’s capacity to understand, negotiate and advocate on AI from worker’s perspectives.
We sought to foster collaboration between education unions and other actors, and to gain clarity on key union issues on AI. Education was selected given its high political salience, contested policy environment and momentum already underway from education unions.
By the end of the programme, the unions had produced a joint public statement on AI in education, establishing a common platform that fed directly into government consultations and external engagement. Individual officers reported significant gains in knowledge and confidence. Several unions restructured internally as a direct result, creating dedicated AI leads and breaking down siloes between policy and workplace-organising focused teams. All nine ended the project with substantive work underway.
Along with articulating specific challenges for unions in the public sector, we identified three considerations for building union capacity on AI. First, programme design involves a trade-off between breadth and depth. For example a deep focus on collective bargaining produces momentum, while a wider spread of issues helps unions develop a more comprehensive approach but with less immediate impact. Second, whereas a diversity of participating union roles can add a range of perspectives it can be tricky to translate discussion into shared goals and actions. Most importantly, the programme affirmed that a context-driven approach to AI can help unions scope out the noise that surrounds the topic and focus work on the specific issues that AI presents for their members.
Introduction
Fair and effective AI adoption – and whether this should happen at all – requires workers to have a stake, from individual workplaces to the wider economy. Now, workers and unions are figuring out how to give that principle force. What is it that workers actually want? What expertise and organisation is needed to get there?
That's the backdrop for a collaborative project between the TUC and Connected by Data to work with unions to catalyse capacity to organise, negotiate and advocate on AI.
This report summarises part of that work: an action learning project involving nine education unions over a seven month period.
Our goals
1. Foster collaboration between education unions
2. Establish clarity on key union issues on technology and AI in education
3. Enable union-led development of a proactive bargaining strategy
4. Build sustainable links between unions, civil society organisations and academics
5. Advance education union influence on technology and education policy issues
Longer term: Inform the TUC and Connected by Data’s ongoing work to support unions to understand, negotiate and advocate on AI.
Our assumptions
The challenge of developing union capacity around AI could be approached from many angles. To help us through, we made a few key assumptions informed by previous work and by conversations with participating trade unionists.
1. Context is key
There are overarching concerns about AI for all workers. However, union responses are more effective when they're tailored to reflect a specific context. Workers from different industries experience different technology applications, impacts, regional regulatory and professional regimes, and ethical concerns. Simply put, what logistics workers need is quite different to what civil servants or computer programmers need.
2. Recognising the particular power of public sector workers
Public sector workers occupy a particular position of power on AI adoption. They are subject to AI technologies as workers, but they are also deployers of AI with wide-ranging impacts on citizens – whether students, patients, benefit claimants or planning applicants. Public sector workers are well-organised in unions, granting them leverage in the workplace: 49.9% of workers in the public sector are unionised relative to the private sector at 11.7%. Public sector unions are also explicitly champions of public services, enabling them to form coalitions and campaigns with wider communities.
3. Create capacity, not outputs
There are many policy ideas and reports. There have been many events and conferences. But we don't yet have the capacity to build and sustain the momentum that makes negotiations powerful. We set out to build capacity rather than create outputs. There is a need to foster relationships between unions and civil society, create ongoing collaboration and deepen the foundational expertise that can confidently keep up to date with the fast-changing landscape and workers and union needs.
4. A learning exercise to inform next steps
There is a stark power imbalance between workers and employers, governments and tech companies. Redressing that is key to having more purposeful, fair and effective technology and policy. As unions and allies we need to learn better how to achieve this rebalancing. This project is a learning exercise to inform next steps.
What we did
A focus on education
Though education is a devolved policy area, the drive from the UK Government – and the enthusiasm of some employers and educators – to rapidly roll out AI in education throws up sticky questions of pedagogy, practicality and professionalism across all age groups and learning requirements, including SEND. Yet too often the voices of support staff, teachers, school leaders and education specialists and their unions are squeezed out. In a period of stress for the education system, there is an appeal of the promises of EdTech products offering to solve everything from staff workload to student attainment and attendance. Education unions had been self organising to build responses to this, and sought collaboration with TUC to help them progress in the interests of members.
The millions of educators across schools, colleges and universities retain deep, embodied knowledge and experience of the realities of education and their student’s needs. They are the group most trusted by parents to make decisions about the use of AI in education, well above EdTech companies and the government, according to a TUC commissioned poll. Finally, education is a politically salient issue for which public trust is key. This gives education workers leverage and opportunity to make common cause with others.
Education unions already engage with governments through various mechanisms. This includes in England the high level Improving Education Together forum, bringing together unions, employers and government in a social partnership structure. In Scotland, the International Summit of the Teaching Profession, has provided a platform for bilateral discussion between the Scottish Government and education unions. However – especially in England – education is fragmented across local authority maintained schools and state funded academies, in addition to a range of independent schools, early years, further and higher education institutions and specialist provision such as in prisons. This further emphasises the need for a contextual approach to interventions by workers and their unions.
A cross union effort
18 participants from 9 unions participated: UNISON, Unite, UCU, NEU, NASUWT, NAHT, GMB, EIS, and AEP. Together, these unions represent hundreds of thousands of teachers and lecturers; school and college leadership; teaching assistants and support workers; early year practitioners; educational psychologists, across the UK.
Participants were national officers responsible for supporting union representatives and members, as well as policy officers responsible for the union’s positions on AI and other areas of policy.
A seven month action-learning approach
Action-learning is the process of bringing thinking and action together. From September 2025 to April 2026, participants met monthly, online or in person, to engage with a core concept on AI and education. Sessions were supported by guest speakers from academia, civil society, the private sector and unions abroad. They then worked to actively plot actions and next steps, together or independently. The workshop programme was shaped by participant interests and priorities, while we curated preparatory readings and external speakers.
Monthly action learning workshops
Workshop 1: Strengths and weaknesses of the movement. Discussing competing visions for AI and education and sharing personal perspectives.
Workshop 2: The policy space for education and AI, with guest Professor Rose Luckin. Exploring potential joint union actions and advocacy.
Workshop 3: Private sector tech actors, with guest David Railton, Director of Education and Public Services at Faculty. Collective development of a joint union statement on AI.
Workshop 4: Strategies and practicalities for collective bargaining on AI, with guests Oxford University Professor Rebecca Eynon and American Federation of Teachers activist Eric Rader. Planning union actions on bargaining.
Workshop 5: School and trust-level AI policy, with EdTech Lead at Chiltern Learning Trust and adviser to the Department for Education Christian Turton. Next steps on bargaining and sustaining collaboration beyond the project.
Workshop 6: The AI aspects of the Schools White Paper, with Educate Venture’s Dina Foster. Aligning unions' responses to the White Paper.
Workshop 7: Collective evaluation of the project. Deciding practicalities for continuation of the group.
What came out of the education unions project?
We sought feedback from union participants throughout the project, including an initial survey at the start; one-to-one interviews after the third workshop; and a final feedback workshop and questionnaire at the end.
1. A public and unified foundational position
The most concrete output was a joint statement – An urgent call for educator voice in AI and EdTech – signed by participating unions. It set out the politics, policy and practicalities of how “educator voice must be empowered from end to end, from the highest levels of regulation to decisions at local levels”.
In anonymous feedback, several participants described it as a clarifying moment: "the joint statement provided a beneficial platform to support our calls for greater educator and representative voice in the implementation of AI."
The joint statement gave individual unions a common platform to anchor their external engagement, informed work with the Department for Education (DfE), and fed directly into responses to government consultations. As one participant put it: "the joint statement has been very useful to inform our external engagement work and internal discussions."
The joint statement has since formed the basis of several union submissions to government policy inquiries or calls for evidence.
2. Individual confidence and knowledge
Participant feedback was consistently positive on the impact of the project on individual knowledge and confidence. The most common theme was the value of exposure to other unions' approaches and to external expertise:
"The pool of knowledge in the group was considerable, and it was useful to share and compare experiences, contexts and challenges. It was empowering to recognise common challenges and opportunities faced."
"The provocations provided by third parties were highly effective in situating the TU [trade union] approach within a wider aspect and in reflecting on how to devise an effective TU position."
"Taking part in the project has deepened my knowledge of AI in education, both in terms of practical knowledge of definitions and tools, and of impacts I hadn't considered before."
3. Internal momentum and collaboration within unions
As well as the statement and the shift in individual knowledge, the project produced internal momentum within unions to clarify roles, remits and structures on responding to AI.
"the issue of who 'owns' the AI remit within the union is one we must reflect on further, internally." Another made the structural point that unions need to "identify people in policy roles and ensure they have AI as a constant/continuing thread in their work".
"As a consequence of joining this project, I have become the AI lead in the Education and Children's Services team, where before the team did not have a dedicated AI lead."
"The project has reduced the siloed nature of development that was present previously by bringing together expertise from across the union directorates that cover a range of responsibilities."
"This has highlighted the importance of working across the union's departments, and the links for example between equalities, education policy and employment relations."
One respondent noted that working alongside policy colleagues pushed industrial officers to think in ways they otherwise might not: "it complements and gets us industrialists into other ways of thinking."
Another participant described drafting updated principles on AI in education for submission to their Education Committee, noting these would "provide the basis for further work to support a bargaining toolkit."
4. Political perspectives on AI diverge
AI is a contentious issue, and there were diverse views on AI within the group alongside shared recognition of risks and opportunities. One argued that "the TU movement needs to unpick arguments around its use for effective public sector reform and the idea that [AI] can drive positive economic development".
Another participant was direct about the limits of what the TUC can drive: "I think there is probably a lack of leadership within the unions that is needed to drive forward a common political agenda on the issue. That's not something that can be solved easily and should not necessarily be driven by the TUC itself."
All unions are now taking significant practical next steps
All nine unions ended the project with a clear set of actions and initiatives now underway prompted by this work – from new policy interventions and negotiation of agreements with governments, through to development of guidance for negotiators and training materials for workplace organisers.
Importantly, most unions are iterating their organisational practices to maintain momentum, including creating new roles and internal structures to break down siloes.
What we learned
About union responses to public sector adoption of AI
Our work with education unions highlighted several challenges that public services, and public sector workers, are going to need to face as they adopt AI.
How to navigate AI-driven public sector transformation
While AI is not something that can solve all the problems within public services, it will change them. Some of this is driven by changing demands brought about by the public’s own use of AI, some by policies that push AI adoption. Both require substantial organisational changes that impact all public sector workers, including those who are using common administrative tools. Unions, reps and members are uniquely placed to provide on-the-ground insight into the changes that are needed; co-design specialist products and services, and the policies surrounding the adoption of general-purpose tools; and to advocate for a long-term view, including how to mitigate the impacts of deskilling on the existing workforce, and maintain a pipeline of future workers.
How to preserve what matters in public services
AI products and services implicitly encode values and political choices. In education, student-facing EdTech can only measure certain things, such as time taken on tasks, percentage accuracy on questions, or when homework (or preparation) is started and completed. But without a professional educator it is hard to measure conceptual comprehension or identify the missing piece that's blocking understanding in a student; hard to measure their critical thinking or creative skills; how happy they are, how frustrated, how fulfilled, or how much they trust the adults in their lives. When these things aren't measured, they may stop official recognition, but they still matter to people, and this disconnect impacts public attitudes to the state. Public sector workers and unions, as well as academics and other civil society organisations, are vital advocates for nuanced perspectives on what good looks like, the role and approach of the state, and its relationship with the people it serves.
How to ensure AI impacts are equitable
There are big disparities between how AI is adopted in different settings, such as across different schools and learning communities. Some hope that AI can help under-resourced public bodies catch up; the reality is that they are likely to struggle most with AI adoption, either not adopting AI at all, or adopting it in ways that are more likely to harm, and less likely to help, citizens and staff. Even within organisations, individual workers can have very different attitudes to AI, ranging from early and enthusiastic adopters, trying it out even in places where it hasn’t been proven, to those who are deeply sceptical, believing that AI is immoral, actively damaging and should be resisted. Unions have to manage this diversity, balancing professional agency and autonomy (both to adopt fast and to not adopt at all); the need to innovate and learn-by-doing; rules and safeguards; and training and support.
About designing union capacity-building projects
Breadth versus depth
We targeted catalytic support for a small group in a single sector across seven sessions. The theory was that deep support for a small number of individuals would have wider effects within and between their unions.
This small group of participants was matched with sessions that ranged across a wide variety of topics. While this enabled engagement with a spread of issues, it limited depth on any particular topic. For example, we weren’t able to bottom out a full common bargaining approach, especially given the different job groups and education sub-sectors represented. We did however see progress made by each union, which has been rooted in the joint statement and informed by the programme overall.
Future projects might either narrow the topic range to allow more sustained engagement per issue, or extend the timeline to allow longer term work on priority areas.
Target participants and programme design
Mixing national officers with industrial, member engagement, and policy responsibilities generated productive exchanges of perspective and having two participants from each union helped manage absences and created internal collaboration. When moving from discussion to action, the different day-to-day demands of those roles pulled in different directions, and a more uniform participant group would allow more tightly tailored programme design, but the cross-function dynamic was also one of the project's strengths. Future projects could attempt to focus on activities that bring different aspects of union work together, such as bringing policy principles into bargaining positions; or shaping member engagement surveys that inform or back up policy work.
Union role design and resourcing
Given massive competing demands, most unions do not currently have roles dedicated to AI as a policy or industrial issue. Responsibility tends to be folded into a wider brief, sometimes informally picked up by individuals with a particular interest. In many ways this is suitable for ensuring AI expertise is incorporated into core issues and expertise. However, limited formal allocation of AI within roles did create pressure from other priorities. As AI becomes more of a focus in various ways, unions may want to consider how to reflect that within job roles, workplans and internal coordination.
Action learning as an approach
The action learning methodology was broadly effective at creating an expectation of active engagement rather than passive reception. Structured activities using online whiteboards supported collective problem solving and record keeping in ways that a conventional seminar format would not have. The month gap between sessions was intended to give participants time for the delivery of actions.
Two issues inhibited more tangible outcomes within the project timeline. First, the range of topics meant sessions moved on quickly, limiting deeper engagement with individual issues. Second, given other responsibilities, follow-through on actions was often limited in scale. We did not include hard accountability mechanisms in sessions such as requiring – rather than merely encouraging – updates on actions. Both are addressable in future design: fewer topics, more time per issue, and more explicit accountability mechanisms between sessions.
What comes next
Given AI’s emergent nature, this work will also have to be responsive to changing technological, social and institutional dynamics. In the first instance however, we have made significant progress. By dropping down from the abstractions and high-level principles into how we actually build the power to contest AI, all participating unions have made strides in individual confidence, organisational infrastructure and practical next steps. These are the prerequisites of an effective movement to give force to the principle of worker voice on AI.
Importantly, the group intends to continue meeting, and two unions have volunteered to chair future sessions with the first scheduled at time of writing. The immediate priorities identified by participants are more bargaining-focused work, the development of joint positions on procurement and commercialisation, and the creation of practical tools that reps can use - Including a TUC supported policy.
The TUC is now exploring how to scale, replicate or iterate on this work towards building capacity within the wider movement. This will include potential new cohorts for action learning, training development and targeted development of union staff, seeking to bring together policy, campaigns and union education into a coherent project. The outcomes of these initial explorations will shape the second phase of this project.
Connected by Data is building on this with campaigning to ensure that the voice and interests of education and general public sector workers are at the centre of official plans to accelerate AI adoption and to bridge the gap between civil society, unions and government officials and decision makers.
Acknowledgements
Thanks to the Ford Foundation for supporting Connected by Data's efforts to empower public sector workers to shape use of AI.
Stay Updated
Want to hear about our latest news and blogs?
Sign up now to get it straight to your inbox
