Don’t be fooled: the idea of using artificial intelligence to assess and predict employee performance is nothing new.

However, the question of whether to use AI in lieu of traditional human-to-human performance reviews and development chats continues to bring up new questions to this day.

As early as 2018, IBM claimed that its AI system, Watson, could predict employee performance with 96% accuracy. It could also determine which worker would become a flight risk with a likelihood of quitting.

IBM’s HR solution offered a “predictive attrition” tool that was ahead of its time. The only downside back then was the fact that, despite Watson’s ability to analyse multiple data points to arrive at its conclusions, “it took time to convince company management it was accurate,” then-IBM CEO Ginni Rometty said.

Fast forward several years. Now, you have 75% of workers saying they prefer AI performance reviews to traditional feedback from a human manager, according to findings from enterprise tech firm ServiceNow.

“This data, as odd as it sounds, is focused on the topic of trust,” said HR industry expert Josh Bersin, who commented on the results.

“People know that managers are biased, so any HR-delivered performance review, pay increase, or other feedback is likely to be biased in some way,” he said.

“AI, on the other hand, has no ‘opinions’. When implemented on real data, it’s likely to be more ‘trusted’ – and 65% of respondents were confident that AI-based tools would be used fairly.”

For Josh, this trust in AI models stems from the explainability of the data and outcomes.

“You can ask the AI ‘why did you select this candidate,’ or ‘why did you rate this employee in this way,’ and it will spit out a precise and accurate answer,” he explained.

People, meanwhile, “have a hard time explaining their decisions”.

Reducing bias in performance management

But would AI-based performance reviews effectively reduce the influence of bias on recommendations?

For Sascha Eder, CEO of market research firm NewtonX, AI can correct for racial and gender bias.

“It also is not susceptible to performance review specific biases, such as recency bias, where actions performed recently are given more weight than actions that occurred say, 11 months ago for a yearly assessment,” Eder said.

“Similarly, AI can control for contrast bias, which occurs when a manager compares an employee’s performance to their peers rather than to objective measures of success.”

Eder, however, is aware of the pitfalls of trusting AI blindly: “That said, AI algorithms are only as good as the training data they are fed.”

Garbage in, garbage out, as AI training experts say.

“AI doesn’t suffer from logical fallibility as long as the data it’s given doesn’t either,” Eder said.

Real-time feedback: Continuous data collection and analysis

The benefits of using AI in performance reviews doesn’t stop at reducing bias.

For decades, most methods of performance reviews have relied on the employee and manager evaluating events based primarily on their own recollection.

“Traditional appraisal methods, though effective for basic evaluations, often fall short of capturing the interpersonal skills and the growth potential of employees,” said Al Montather Rassoul, CEO of MENA-based consulting group, MRC Firm.

Rassoul believes integrating AI along with emotional intelligence into performance reviews is “revolutionary” in that the AI model will assess both quantitative performance metrics and the qualitative attributes unique to the individual under evaluation.

Through automated data collection and analysis, the “AI monitors metrics like task completion rates, response times, and overall productivity,” he said.

Managers, therefore, have real-time data on employee performance.