If you feed an AI your company's promotion data from the last 10 years, will it predict future success, or just replicate past biases?

Be careful: AI trained on biased data doesn’t reduce risk — it multiplies it.

A powerful example is COMPAS, the algorithm used in the U.S. justice system to predict reoffending risk. It was meant to be fair and data-driven. Instead, it consistently flagged Black defendants as “high risk” at nearly twice the rate of white defendants.

The algorithm didn’t find criminal patterns — it found patterns of systemic injustice and called them “truth.”

The reason this happens is simple: an AI optimizes to find patterns, not to be fair. If you give it data reflecting systemic injustice, the AI won't correct it; it will learn it, scale it, and call it "efficiency."

At Board&Leaders, we’ve chosen a different path.

We eliminate this problem at the source by not relying on subjective historical data. Instead, our proprietary AI operates on a validated, bias-resistant framework rooted in neuroscience: the Consistency model developed by Klaus Grawe. We combine this foundation with objective biometrics and cognitive responses to assess potential independently of past performance or evaluator bias.

This allows us to understand a person’s potential with no dependency on an interviewer's opinion or flawed historical records.

In the end, the COMPAS case teaches us a fundamental lesson. In human decisions, the quality of your data isn't just a technical issue—it's a moral imperative.

And we chose to start from a clean source.

Learn more about the COMPAS case and the risk of multiplying bias through AI:

Machine Bias” – ProPublica (2016): ProPublica found that Black defendants were nearly twice as likely to be incorrectly flagged as high risk compared to white defendants.

AI Bias: How It Impacts AI Systems (2025): Tredence explains how AI bias can lead to unfair hiring, lending, and customer engagement decisions. Explore the types of bias in AI systems, real-world examples, and strategies to build ethical, unbiased AI.

Serious Games and Virtual Reality for Mental Health (2021): A Review of Recent Developments (Giglioli et al. (2021): Virtual behavioural simulations grounded in the Consistency Model offer a low-bias alternative to traditional tools by providing more ecologically valid environments and directly measuring responses to real-world psychological triggers.