by Kathryn McKenna
Linda Hynes, Employment Law Partner in the Employment, Immigration and Reward division based in the Dublin office at Lewis Silkin, has warned of the discrimination risks surrounding emerging AI technology.
Linda spoke with Sync NI as concerns over discrimination risks with new artificial intelligence technology have emerged with possible implications to employers, specifically throughout the hiring and onboarding process in recruitment. It follows recent news on AI innovation and technology - with facial recognition software in particular making headlines recently.
Questions have been raised as to whether AI poses an increased risk for discrimination claims, with a delivery provider recently facing two separate race discrimination claims in the UK. Whilst in Ireland, a case was heard by the Workplace Relations Committee in 2022 relating to whether a job applicant, who received an automated rejection by email, had been the subject of racial discrimination.
In a recent report by the Workplace Relations Commission, it was found that over 40 per cent of HR departments use AI applications. The report also references the use of the AI Act in Ireland, and how EU implementation of this will decrease discriminatory outcomes impacting worker’s careers, livelihoods, and rights.
Although not regulated as of yet, Linda believes that employers should ensure to use best practice when using AI tools in the job selection process.
“There is a lot of talk around the EU AI Act and I think it is going to have significant implications in the HR and employment space,” Linda explains. “We are already seeing a lot of employers using AI tools, particularly around recruitment and selection so employers are going to have to consider the implications of the EU AI Act.”
Linda also cautions that employers using AI for recruitment or performance management should ensure AI-supported decisions are explainable and transparent. Linda said: “Employers might think that if they are not the developers of the system it doesn’t impact them because they are just using a system.
“But ultimately, if there are potential uses that are contrary to data protection and employment equality obligations, those issues will sit with the employer. If the AI tool is inadvertently discriminating against a candidate for a role, that does not mean that the employer will not have to deal with the equality case which might arise from that.”
READ MORE: 'Having low numbers of females in STEM is a repetitive cycle that needs to be dealt with'
Linda adds: “I can see challenges for employers trying to explain what decisions have been made by these tools and having to try and defend allegations that the employer’s decision was discriminatory because of their use of particular AI tools. It is really important all employers start thinking about this.”
With the possibility of this calibre of claim increasing, it is essential for employers to retain good practice when using AI in the hiring process.
“Everything is moving so quickly and the regulation is trying to keep up. It is really about sitting back as a company and doing your inventory and asking if there is anything you are currently using that could be covered by the EU AI Act that you need to consider your compliance obligations for.”
Decisions must be monitored for bias, as Linda explains: “Companies need to ask themselves, ‘Is there anything we are using or intend to use that could be potentially making biased decisions that we need to start doing our due diligence on?’”
Linda emphasises how vital it is that employers do their ‘proper due diligence’ on the technology company which is providing the tool to try and understand how the provider tried to deal with the potential for bias in these systems. One of the critical questions employers need to be asking is, ‘what is your technology being trained on?’ If the underlying data sets that are being used have biased decisions within them, then the tool is going to potentially replicate those. If you are using 50 years of recruitment and selection information which has traditionally favoured people from certain gender or ethnic backgrounds, then your tool is going to potentially replicate those biases.”
Linda explains therefore it is key that employers do their research: “The provider, the developer of the tool, is going to have the most obligations under the EU AI Act. But the deployer, which in this scenario is the employer, will also have obligations including making sure the tool is used only in accordance with the instructions. It should only be used in the way it was designed and intended to be used by the provider - which is critical for employers to be mindful of.
“If you are an organisation using these tools you will want to carry out your due diligence to ensure that the company that develops them and provides them to you has done all the things they need to do under the EU AI Act. That includes making sure that the underlying data is as unbiased as possible and that it meets certain quality criteria. Employers themselves also need to ensure they have someone assigned to oversee systems who can deal with issues that arise. Also, where they are using their own data as part of training these tools, they need to make sure those data set have as much integrity and quality as possible.
“There will also be obligations around monitoring and keeping logs around these systems, so transparency is going to be really key.”
However, when used appropriately Linda agrees AI tools certainly have their place, particularly when it comes to upskilling opportunities: “There is some fear as a result of scaremongering and headlines asking, ‘Is AI going to replace us and take away our jobs?’ But actually, I think there are a huge amount of positives with the opportunit to make people’s jobs more interesting and to get rid of more repetitive-type tasks.
“I do not believe AI will replace all roles, but I do think that roles will change and become more strategic and less administrative. There is an opportunity here for people to train themselves in these new technologies so that when they do need to use them in their roles, they will be up-to-speed on the benefits and how best to use them.
“There is a lot of fear around AI, and generally when people are afraid of a topic they try to avoid it. But there are opportunities to skill up on it which should be taken, because AI is going to start being a discussion in every facet of the workplace.”
Subscribe to Sync NI Newsletter for the latest technology news, jobs and upcoming events in N.I.
The latest Sync NI magazine is out now. Read the magazine now online for free here.
