As artificial intelligence becomes embedded in more business processes, organisations are having to confront a question that goes far beyond technology: who remains accountable when an automated system gets something wrong? asks Susie Thomson, pictured, Chair Elect, Professional Background Screening Association (PBSA).
Recruitment provides a particularly interesting test case. AI can now help organisations process vast numbers of applications, identify potential candidates, verify information and streamline screening. These capabilities offer significant efficiencies, but they also introduce a new layer of risk. When technology begins to influence who progresses through a recruitment process, organisations need to be certain that efficiency has not come at the expense of fairness, transparency or accountability.
An ongoing lawsuit in the United States illustrates precisely why this matters. Mobley v. Workday centres on allegations brought by Derek Mobley, an African American man over the age of 40 with a disability, who claims that Workday’s algorithm-based applicant screening technology discriminated against him when he applied for jobs using its recruitment platform.ย The case alleges discrimination on the grounds of race, age and disability, but the wider issue should interest risk, security and governance professionals well beyond the United States. If an automated tool influences or contributes to a potentially discriminatory outcome, where does responsibility ultimately sit?
Whatever the eventual outcome of the case, it raises questions that organisations using AI should already be asking. One of the dangers is assuming that because a decision is supported by technology, it is automatically objective. AI systems rely on data, parameters and processes created by humans. Historical data can contain existing biases and previous patterns of decision-making. Without appropriate controls, there is a risk that technology simply replicates these patterns at greater speed and scale.
That is why I believe human oversight must remain fundamental. At the Professional Background Screening Association (PBSA), we see significant opportunities for technology to increase efficiency and enhance recruitment and screening processes. But increased automation should never mean removing professional judgement from the equation.
AI is exceptionally effective at processing large volumes of information, identifying patterns and supporting administrative workflows. What it cannot necessarily provide is the context and balanced judgement an experienced professional can bring to a decision. This is particularly important in background screening. Technology can help screening providers collect, verify and organise information more efficiently than ever before, but oversight, ongoing development and professional expertise remain paramount. Selecting the appropriate AI tool, determining the questions being asked of it and verifying its output are just some of the areas where human involvement is critical.
Organisations should therefore treat AI in recruitment as a governance issue, not simply a technology or HR issue. They need to understand where AI operates within their processes, who is responsible for overseeing it and how its performance is tested. Meaningful human review should remain part of significant decisions, while tools should be regularly evaluated for bias, consistency and compliance with evolving legislation.
Transparency is equally important. Candidates should understand how their information is being assessed and the role technology plays in that assessment. Being open about these processes helps build trust and demonstrates that an organisation takes responsible decision-making seriously. The potential of AI should not be underestimated. Used well, it can make recruitment and background screening faster, more efficient and more consistent. But organisations cannot outsource accountability to an algorithm.
The Mobley v. Workday case is unlikely to be the last legal challenge involving AI and recruitment so rather than waiting for regulation or litigation to expose weaknesses, organisations should be examining their own safeguards now. The future of security and compliance in recruitment will undoubtedly be AI-enabled but if organisations are to manage the risks that come with it, accountability, oversight and ultimately the final judgement must remain human-led.
