Keeping Redundancy Fair When AI Is At The Table

Employers are facing a new wave of AI assisted challenges in already challenging redundancy exercises. From attacks on selection pools and criteria to detailed requests for further information and lengthy appeals, there's a significant increase in both the complexity and volume of challenges raised by employees. We consider how to respond fairly and proportionately.

As we explored in our recent articles, AI is now frequently being used by employees ingrievance processesand also inlitigation. This brings with it real challenges for employers, who are now having to unpick lengthy, repetitive and seemingly legalistic complaints. Unsurprisingly, redundancy processes are not immune from this trend. We explore the key principles which employers should keep in mind, both in the UK and in Ireland.

How is AI influencing consultation processes?

AI related challenges are emerging in the following areas:

• Selection pools and criteria- Employees are harnessing the analytical power of AI to scrutinise selection pools and criteria and produce lengthy challenges. This might be used to highlight alleged bias or other patterns of unfairness. Employers must then review and respond to these reports (many of which contain inaccuracies), requiring significant time and resources.

• Detailed requests and questions– AI is a fantastic tool for brainstorming. But when put to task by employees during a consultation process, the questions generated may be overwhelming in number and complexity. Responding to these, particularly in a large scale process, risks diverting attention from the main process and can impact on the consultation timetable.

• Consultation meetings- AI supported preparation could change the tone of consultation meetings. Employees are now more likely to come armed with an extensive (and sometimes irrelevant) question list or scripted responses, increasing meeting length and impacting the flow of the discussion/consultation. Also, sometimes the language generated by AI-tools is overly antagonistic. AI generated meeting transcripts may also conflict with employer notes, creating disputes over the record.

• Lengthy appeals– Reflecting our experience with grievance processes, appeals produced by AI tend to be longer, more legalistic and potentially peppered with hallucinations. The additional investigation and fact checking eat up time and management resources. This is likely to have a knock-on effect on the redundancy timelines and could disrupt business planning.

• Data risks- If an employee uses a public AI platform to help produce their redundancy documents, there is a risk that they will input sensitive company data, or sensitive personal data about colleagues, raising data protection and confidentiality risks.

Remember the basics

In our article on grievances, we underlined the importance of remembering the guiding principles in the Acas Code in the UK. Similarly, Employers should be mindful to not let AI 'noise' get in the way of conducting a fair and reasonable redundancy process, adhering to the core principles of fairness.

We explore what employers need to do to avoid a successful unfair dismissal arising from a redundancy situation in detailhere. In summary, there must be:

• a genuine redundancy situation
• adequate warning and consultation of affected employees
• a fair selection process
• consideration of alternative employment
• a fair procedure

When faced with AI overwhelm, which elements of this process are most likely to be impacted?

• Meaningful consultation- Consultation is crucial to a fair redundancy and this means giving employees the opportunity to ask questions and have their say. It's important to try to answer the questions and address the challenges raised, but if AI makes it hard to see the wood for the trees, the key points to cover are:

• Enabling the employee to put forward suggestions about the rationale for the redundancy and how to avoid redundancies
• Commenting on the basis for selection
• Challenge their selection
• Considering alternative employment

• Pre-determined outcome– A common challenge in a redundancy process is the assertion that the outcome was a foregone conclusion and the consultation could not impact the final decision. If AI assisted consultation results in an overwhelming number of questions, or perhaps challenges that are repetitive and overlapping, it's important to tackle this directly. Simply failing to answer the questions without explanation risks exacerbating the claim that the outcome of the consultation process is pre-determined.. In our experience, meeting employees in person is a good way to address the issues and demonstrate a fair process without spending enormous amounts of time engaged in lengthy written responses (see below).

• The timetable- Aside from the minimum length of collective consultation processes (depending on the numbers), the law does not prescribe how a how long consultation should last to be fair. Often employers will have a timeframe in mind, and complications arising from AI challenges could threaten that. Key principles to focus on are that consultation needs to take place at a formative stage, and that timetables should have sufficient flex in them to enable questions to be answered.

• Right of appeal?– The need to submit a written appeal is a less daunting task in the age of Chat GPT and experience suggests that employers now risk being inundated with lengthy challenges. Employers may therefore be asking whether the process must go this far – is an appeal essential in a redundancy process? The answer is that it depends. In the UK, not offering an appeal won't automatically mean that a dismissal is unfair, but it could do. Acas guidance is that this is good practice, and it can of course provide another opportunity to remedy any defects in the process and shore up the process as much as possible. In Ireland, the circumstances where an employer might not offer an appeal are more limited. For example, it might be justified where an entire company/department are closing down, such that an appeal is highly unlikely to overturn the decision to dismiss. A fair process in Ireland requires observation of the constitutional right to a fair process, so in almost all dismissals, a right of appeal should be offered. (Although if an ex-gratia payment is being offered at the time of dismissal in consideration for a waiver/release then this could reduce the numbers who actually appeal.)

Practical strategies

There are some practical strategies that may help employers keep these consultation processes on track:

1. Face-to-face consultations– Engaging face to face and not becoming entrenched in correspondence will be important. This may help clarify the employee's key concerns, minimise AI generated correspondence and safeguard the human element of the process. Meetings in person are ideal, but remote meetings are also acceptable. The key, however, is to try to draw the focus away from lengthy AI-generated correspondence and towards the employee's own perspective. Agreeing a summary of the employee's key points may help to reduce the volume of AI"noise".
2. Transparency and records– Consistency and explainability is the best defence against AI amplified challenges. In practice, this means that, the more you are able to back up your proposals with evidence from the outset, the more confident you will feel about managing the process.
3. Guidance documents– Provide managers with the tools to retain focus on the core requirements of a fair process. Template discussion points for consultation meetings may be helpful, but, managers and HR should be reminded that these meetings must be a genuine two way dialogue.
4. AI usage policies– As noted above, there is a risk that employees' use of AI in this context prejudices company data or that of other employees. To mitigate this, employers should implement clear policies governing the use of AI and provide all staff with training on responsible usage, including what types of data are appropriate to share.
5. Timing of settlement offers– If settlement agreements are on the cards, consider the timing of these. Early settlement offers could be a useful way of resolving potential disputes swiftly, thereby saving time and resources that would otherwise be spent on managing AI-generated challenges.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.