AI in the Workplace: A Brief Intro
Adopting AI as an employer can be complex. What employment law risks are associated with using an AI tool in the workplace and recruitment? How do you navigate these?
It is important to identify not only the risks but also the likelihood of those risks occurring and what the consequences would be. You can then mitigate these risks by, for example, training staff and imposing human review of the AI output.
To begin with, you should try to find out from your supplier whether there could be any unfairness or discrimination in the output. This can be ascertained by requesting information about both the data and method used to train the AI tool, and how it currently performs across people with different protected characteristics. For example, facial recognition software has in the past been less accurate for those with darker skin, which may amount to indirect race discrimination. It is important to find out whether any testing for this has taken place, and whether there have been any previous employment law claims, in relation to the product you are looking to use. If any bias has been uncovered, you should follow up by requesting evidence of any mitigations put in place to deal with it.
One of the most important things to consider is accountability for decisions. You need to not only keep accurate records of decisions made, influenced or impacted by AI, but also the reasons for those. This means understanding how the tool works sufficiently well that can to explain its decisions. You may need to justify yourself (as an employer) to an affected individual or a Tribunal down the line. For this reason, it is important that a human can also intervene to influence the output i.e. approve, modify or override decisions made by the AI tool if an individual raises a grievance about it, for example.
Bear in mind when it comes to ‘affected individuals’, that these may be people who don’t use the AI tool directly, but are nevertheless impacted. For example, if a manager allocates shifts to their team using AI. In this case, you would also want to check that such allocation is permitted under the employment contract.
A couple of other risks which you would need to address include how protected disclosures submitted via AI will be spotted and what adjustments for disabled users may be reasonable. It will be important to have an ongoing point of contact with your supplier for support as you are likely to need issues resolved quickly. You will also need to update your policies.
Ultimately, engaging directly with your staff an communicating transparently should help limit any employee relations issues which may arise from the introduction of this new technology.
To discuss anything above, including the adoption of AI in your business please call Grace Smyth on 01473 611211, email [email protected] or fill out our enquiry form below.
Grace Smyth is a Solicitor in the Employment Team at Barker Gotelee Solicitors.
