Labour Arbitration Sucks – Part 3: Improving Grievance Management with AI
July 9 2025
Incandescent Bulb. Image Source: Pixbay.com
A Credible Source to Assist in the Resolution of Grievances under a Collective Agreement
In labour relations there are many reasons to send disputes to third party neutrals for resolution (i.e. mediators or arbitrators). There may be an honest disagreement with respect to the facts giving rise to the dispute, the parties could simply disagree with each other with respect to the interpretation of the agreement or there may be a patent ambiguity in the contract language that needs to be fixed up. Notwithstanding it is also quite common for matters to be referred on simply because the stakeholders (management and the employees) lack trust in one another and want their “day in court”.
We may refer to this latter set of disputes as being “political” meaning the labour relations experts on both sides of the table (i.e. union and management) may understand how the matter should be resolved but feel their hands are tied because an agreement with the other party might alienate their respective stakeholder group. Hence matters are referred to a neutral party who is seen as a disinterested authority empowered to impose the “right” decision.
This appeal to authority though has devolved to some extent precisely because unions and labour relations managers are consciously aware of this use of the process and exploit it to their advantage. It is believed that the vast majority of grievances referred to arbitration, faced with uncertain outcomes, are settled before arbitration. This is the result of the parties leveraging the perceived threat of the proceedings and a possible undesired outcome to convince the stakeholder that a settlement is a negotiated outcome and therefore a known outcome and therefore a desired outcome.
This behaviour is common because between the final internal step and arbitration we usually introduce new actors – i.e. legal counsel for both parties. Regardless of the actual expertise of the parties (remember union reps and labour relations managers are often subject matter experts equally as qualified as counsel) the reality is that external experts often are seen by stakeholders as being more credible then internal experts. Legal counsel can often convince a stakeholder (eg. management or an employee) to drop a problematic case in instances where the respective representative could not.
Even so it is perhaps surprising to those outside labour relations that lawyers on each side as often as not do not have the leverage to convince stakeholders to drop their position and make attempts to settle a dispute. Thus, in many instances at the beginning of the arbitration process the arbitrator puts on their mediator hat and tries to help the parties resolve the issues.
Commonly this is referred to as directive mediation, (also known as evaluative or advisory mediation). It is a mediation style used by arbitrators whereby they actively provide opinions, suggestions, and potential solutions to the parties involved in a dispute. This approach is distinct from facilitative mediation, where the mediator or arbitrator primarily helps the parties communicate and explore their own solutions. Directive mediation is used in situations where parties are struggling to come up with and reach an agreement on their own and need more guidance. Apart from the merits of this evolved process, there are significant problems with it. Directive mediation is a process that requires the practitioner (i.e. the arbitrator) to have experience, expertise and a particular natural skill that is hard to find. Hence a small group of arbitrators are favoured for their given skill set and dominate the workload in a given jurisdiction. The result is a shortage of preferred arbitrators and the road to arbitration has become a prolonged and expensive process that no longer resembles the intent of the original lawmakers who dreamt up the process as an expeditious alternative to traditional litigation (i.e. lawsuits before the courts).
One neutral source of information regarding Labour Arbitration outcomes that has become the accepted “bible” is Brown and Beatty’s Canadian Labour Arbitration1 which has, for more than forty years, provided a concise and comprehensive record and analysis of the whole body of arbitration developments in Canada. It offers a credible resource anyone involved in labour arbitrations can use to assist in analysing the merits of a given case.
The original author of Brown and Beatty’s Canadian Labour Arbitration book is currently working on an Artificial Intelligence product that will enable users to upload documents related to a given grievance (including the collective agreement, copy of the grievance and all documented supporting evidence, previous related settlements) and receive outputs that include a summary analysis of the facts and the merits of case and potential outcomes. This tool will allow users to analyse grievances and will support better grievance resolution outcomes.
The intention is not to replace the people in the process but to provide objective information that will assist the parties review the facts, develop solutions and arrive at settlements internally. It will provide a third-party service that gives indications of how a matter might be resolved at arbitration and thus informs the respective representatives on likely arbitrated outcomes.
Brown and Beatty's AI product has a long term vision to help parties prepare and assist anyone at any stage of the grievance process, with an ambitious goal to reduce the need for arbitration by advising on what the probable outcome would be. AI is not a panacea that replaces human interaction in the labour relations context by any stretch of the imagination. However, Brown and Beatty have found a way that allows labour relations practitioners (union, management and third parties included) to use modern tools to improve existing processes and develop better outcomes.
Another by-product of the AI tool is that as it develops and more users in the labour relations community begin to use it we will be able to develop better understandings of grievance-arbitration metrics. As well, while there is a great body of labour jurisprudence created from reported cases (Library and Archives of Canada) and the wide use of AI may help build an even larger data base that will enhance the reliability of any analysis and contribute to a more thorough understanding of why certain decisions are made with respect to labour relations issues.