Tackling unconscious bias: is training the answer?

Don’t buy unconscious bias training. Don’t run it. Don’t spend a penny on it. Laugh anyone out of the room who’s trying to sell you it. Yep, even you Keir Starmer. I’m not saying this as a belated dig at the Labour leader, nor because I don’t believe in equality.

It’s because unconscious bias training doesn’t work. And the real problem is, it looks like it should work. It relies on the mantra ‘if you can shine a light on a negative thought, then that’s the first step to changing how you act’. Which is logical, but wrong.

Modern behavioural science knows our behaviour is rarely linked to our attitudes, and varies wildly due to context. So it’s a hard ask to get massive change from hosting a mandatory employee training session for three hours.

Why doesn’t it work?

Firstly, it’s easy to ignore. Think of a conscious bias you have, like your love for a certain sports team. Let’s say you love Tottenham Hotspur FC (I’ve probably just offended a bunch of people). You’re very aware of your strong affinity to Spurs. Now, love Arsenal FC instead. I know it’s sacrilegious, but try.

There are more Arsenal fans than Spurs fans in the world; so you’ll be correct in a societal sense to support Arsenal. No? Well, how about how much they win? Arsenal have won more than Tottenham over their history. That should be convincing right? Oh, it’s not? Well I’ve probably failed to convince you to swap then. This probably has something to do with me as the messenger not really understanding why you love Tottenham and you being very within your own justification to ignore me. Morally, I’m not you and really you don’t need to answer to me. Nor any trainer or room of colleagues. If our conscious biases can be so hard to shift, our unconscious ones must be even harder. We don’t know they’re there but they’re every bit who we are.

Secondly, awareness isn’t behaviour. Think of an unconscious bias selection you make every day – who you’re attracted to. For the sake of argument let’s say you really like green-eyed people. And up until just now, you didn’t know you did. Now I’m telling you it’s bad to like people with green eyes.

Does knowing that change you fancying people with green eyes? Do you think next time you’re walking down the street you’ll remember me saying that it’s wrong who you like, and you’ll cross the street when beautiful green eyes come your way? Probably not. You’ll forget me and my writing pretty quickly with no change in your behaviour.

And thirdly, it focuses on people rather than systems. Unconscious bias training looks to make your unconscious bias conscious, then, using classic stick and carrot techniques, beat or treat you out of your now conscious bias.

It tries to change individuals in the hope that when they build systems, they’ll do so with their biases in check. Therefore, the systems will have perfect equality. Unfortunately, even in the case of Google’s AI systems, which has fewer human hands in it than most, that doesn’t work.

What we should do instead

The behavioural design process is great at making low effort, low cost, real changes. It looks to work on a granular, idiosyncratic level by changing the systems people work in, rather than the people themselves. Targeting those small behaviours you really want to change with a view of snowballing into better behaviours.

For example, if you want to get more equality in your hiring process, you could use blind CVs, make sure your hiring panel grades interviewees alone (and without discussing their views), and make shortlists that contain diversity. Better yet, use structured real job-like tests instead of relying on interviews. All these measures have been proven to help with diversity and inclusion.

You could couple that with reframing the language of diversity to be about better creativity and more productive workforces rather than equality and being morally right. Again, that’s shown to help.

Like most issues in the world, sitting about for an afternoon, making awkward chat about who we are and what we believe (or didn’t know we believe) in an effort to change behaviour is about as useful as a poster on the wall that reads ‘we value diversity and inclusion’.