I’ve had the good fortune of spending time with some researchers who want to know what I think about personality assessments at work.

Why me? Well, because I see a lot of HR products and platforms. They wanted to know if I think technology can predict good hires from bad ones. Can algorithms and assessments tell managers something about a candidate that will predict them from making mistakes in the selection process?

It’s essential to understand why people want predictive tools to help us understand the thoughts, behavior, and personality of our workforce.

Recruiting is expensive. Turnover is costly. As long as we’ve been hiring, we’ve been trying to predict if an employee will earn back what we’ve spent on hiring them and more. Could this individual make us a lot of money? Will this person hurt our chances of success? How do we know we’re making the correct decision?

Here’s what these models get right: it’s easy and maybe even helpful to classify and arrange your applicants, candidates, or employees into categories of behavior when you only see your workforce within that system. Are they introverts or extroverts? Risk-takers or risk-avoidant? Feelers or thinkers? Capricorns or Leos? No assessment has ever been shown to nail the human heart or mind completely, but some do a halfway decent job of slowing down the hiring process to allow for introspection, examination, and conversation. You get a much more thoughtful understanding of the human being before you when you get people talking about hopes, dreams, and values.

What these models get wrong: most of these systems are biased and tell us more about the people who developed it instead of the people who use it. From Meyers-Briggs to micro-expressions, it starts with a researcher with an idea who brings her inclinations, preferences, and leanings into the process. Then it goes to an employer who has a predetermined view of “culture” that is often just a set of preferences and choices from the people in charge. Then it’s used by hiring managers and internal leaders who are trying to de-risk human activities (like hiring and maybe even talent management programs) without fully understanding that no action with human beings can be fully de-risked.

Personality assessments capture who the applicant, candidate, or employee is today — maybe. What they cannot fully capture is who this person will be once she’s been tested by the events of today. And no matter what anybody tells you, we cannot capture or estimate someone’s capacity to surprise, amaze or even shock us tomorrow.

Sure, we can guess. Sometimes we’ll get it right. We might be able to identify someone who is historically negative, adversarial, aggressive, or avoidant. If we’ve done our homework, we might be able to map out how those past behaviors inform someone’s chance of success in a work environment. But that’s a stretch. Most of us don’t know how and why people are successful at our companies. Or we do know — and it’s because of behaviors and choices we’d rather not admit on paper.

I think the beautiful thing about humanity — and even this great American experiment — is that we have the power to wake up and experience the circumstances of today, and fundamentally change the course of our lives tomorrow. Who we were yesterday doesn’t have to be who we are today or tomorrow. Step by step, brick by brick, we can rethink our choices and rebuild our lives.

What personality assessment predicts change? Which selection model nails optimism and transformation? What algorithm zeroes in on the potential of the human heart?

(Please don’t tell me it’s yours because it isn’t. )

I believe that personalities are both simple and complex, clear and murky, precognated but incalculable. When employers play amateur psychologists, they are as wrong as any therapist who tries to score your life after one or two sessions. And when employers try to know you before you even know or reveal yourself? They’re playing god, which is both illegal and immoral.

How do we know if an applicant, candidate, or employee is the one who undoubtedly changes the course of our company’s history? Are we about to hire someone who brings our organization one step closer to its goals? Can we tell if we’re wasting our time?

I don’t blame organizations for trying to get smarter about HR and make better predictions about humankind. But a little humility helps. The divorce rate in America is over 50%. Over 90% of start-ups fail. People are bad at seeing the future. If you think an algorithm and facial recognition software will help you beat the odds — and beat failure — you don’t know a thing about humanity after all. Personality assessments are dangerous in your hands. And that’s what keeps me up at night.