Metropolitan Police leaders at Scotland Yard have admitted using AI to monitor staff behavior and performance, prompting criticism from the Police Federation union.
After previously denying the use of artificial intelligence tools provided by Palantir, Metropolitan Broadcasting has confirmed that it uses Palantir’s AI to analyze internal data on sickness levels, absenteeism and overtime patterns. guardian It was reported on Sunday.
Representatives from the Police Federation warned that relying on insights from AI could lead to a false interpretation that “unsustainable workload pressures, illness and overtime are indicators of misconduct”. But the force argued that AI tools are needed to improve standards and culture, as evidence from AI tools suggests a correlation between increased sickness, absenteeism and overtime and “deficiencies in standards, culture and behavior.”
The federation reported. guardian: “Systems that use algorithmic patterns to create profiles of police officers must be treated with extreme caution. If the military is serious about raising standards and public trust, it must continue to focus on proper oversight, fair process, and human judgment, rather than automation of suspicion.”
To ensure that AI risk flags are interpreted fairly, business leaders should treat every flag as a conversation starter rather than a verdict, advised AI consultant Colette Mason of Clever Clogs AI. Human resources department magazine.
read more: Leaders have a duty to guide the correct use of AI
“AI needs to find patterns and connect the dots,” Mason added. “The human mind and common sense should make the final interpretation.”
Ellie Smith, head of European research at global non-profit organization Catalyst, agreed, saying: Human resources department The magazine argues that employers should treat AI risk flags as decision support rather than the final decision.
The results of a joint poll released yesterday (24 February) by Catalyst and global think tank Coqual suggest that UK businesses are ahead of their European counterparts in taking steps to ensure that AI data is used and interpreted fairly and that its output does not create or reflect inequality.
Almost half (46%) of UK employees say their organization is taking steps to ensure that AI does not exacerbate inequalities, compared to 34% in France and 31% in Germany.
54% of UK employees said their organization is not taking steps to prevent AI from exacerbating workforce inequalities, lower than in Germany (69%) and France (66%).
Smith suggested that UK leaders should focus more than ever on ensuring that AI-powered decision-making is fair. “Organizations should ask how leaders, HR, and teams across the business can enable equity and inclusion through AI. This is not just an HR issue. Leaders need to set the standard.”
read more: Is your AI fair? How HR can combat algorithmic gender bias
“Only 55% of employees trust their organization to handle AI responsibly,” she continued. “Organizations need a clear requirement for human involvement, which means training administrators to interrogate, contextualize, and override AI output when necessary, while tracking downstream outcomes.”
Mason agreed that manager training is key to ensuring fair and responsible decision-making with AI. “Managers need to understand three things: what AI can see, what it can’t see, and what questions they need to ask before acting on its output,” Mason said.
She added, “Technology discovers patterns. People provide context.”
Smith continued, “Managers need training that builds not just tool proficiency, but also AI fluency, human-centered judgment, and inclusive leadership. Many leaders are unsure of how AI systems work and where bias and errors occur.” “Training should focus on how AI models generate outputs, when human review is required, and how leaders apply judgment, accountability, and confidence in the face of uncertainty.”
Catalyst and Coqual surveyed 2,891 business leaders and employees in France (34%), Germany (31%), and the United Kingdom (35%) and turned to IBM SPSS Statistics to analyze the results. Each organization also gathered qualitative insights from 31 individuals who did not participate in the study. This includes executives, functional leaders, team leaders, and experts in AI, talent management, IT, and data science from multinational companies headquartered in Europe.
