Tech Hiring and Dancing Chickens
Past Performance != Good Predictor
Ever heard of the Amazon recruitment AI system that had to be shut down because... it was too biased [^1] ?! Yes, this actually happened!
Making good hiring decisions is hard. Everyone recognizes that. Even so, few attempt to do anything about it. Fewer at the scale of Amazon. They looked at what resumes led to 'good' hires and tried to build a winning-horse spotter. They trained their AI model on 'good' resumes in order to identify potential good hires and to weed out not-so-good ones.
The trouble is, all those 'good' resumes belonged to men. And the AI model inferred that women, therefore, are unlikely to be good hires!
Notice anything strange?
As seen in the popular selective attention test [^2], we have a tendency to hyperfocus and miss the bigger picture. This isn't to malign Amazon or indeed anyone who attempted to build this system. On the contrary, my intent is to shine a light on a much more basic problem: no wonder the system came to that conclusion. The overwhelming majority of our new hires are men!
Not correcting this has a cost.
I've interviewed and hired hundreds of engineers in the past. Hundreds if not thousands (stopped counting after 2011 - 92 interviews across the year, 58 in the first quarter alone. Hired two. Yes, two men.).
I'm disappointed I haven't critiqued my biases earlier and developed ways of counteracting or minimizing them. I remember several occasions where I've decided to hire in the first five minutes of an interview and spent the remaining time confirming what I believed to be true about the candidate (confirmation bias). I've been prejudiced when looking at CVs with gaps or lacking university education credentials. I've interviewed candidates with a really narrow view of 'expertise'. I haven't taken a leap of faith enough times with women candidates as I have with men. The list goes on.
And I'm not alone in failing to do so either. We all know this is a significant problem for the entire tech industry. And we fail, as individuals and organizations.
So how do we hire better?
This is indeed the question we need to tackle. Unfortunately, hiring as a capability is really underestimated in most firms hiring tech folks. As such, no formal models exist whereby firms can embed their values into their hiring practices and continually improve them.
Relax, all hope is not lost yet. There are a few emerging trends which we could potentially build upon. Here are my top three:
Trend #1: Investing in returner programs
And so we return to the same firm that we started the article with. Amazon in 2021 expanded its return-to-work initiative by hiring 1000 individuals. Most of the top tech firms (Facebook, Apple, IBM, etc.) have active programs. Harvard Business Review recently ran an article that claimed roughly 16 of the Fortune 50 companies had an in-house returnship program for mid-career professionals[^3]. In spite of the recent hiring downturn, the battle for talent shows no signs of abating. Returnship programs offer an excellent funnel to recruit and grow skilled professionals.
Trend #2: Engaging with recruiters focusing on diversity
It's quite all right if you're one of the top 50 firms worldwide. Startups, SMBs and firms with a smaller footprint in some geographic regions struggle to hire and retain top talent. This is where focusing on career returners can offer that much-required fillip. Those projects won't get done if we limit our talent pool to current or ex-FAANG male employees. We need to widen the net. Working with recruiters focussing specifically on diverse candidates is another growing trend.
Trend #3: Curating job descriptions to attract diverse candidates
You've probably heard this factoid before: Men tend to apply for jobs even if they only meet 60% of requirements in the job description. Women, on the other hand, tend to apply only if they match 100%.
There are some low-hanging fruit that one can immediately pick to shift this statistic. Clear requirements in the job description, ensure must-haves are absolute must-haves, using the right language, and reiterating your flexibility/diversity credentials are a few strategies that have yielded very good results for most. This is an excellent place to start.
Final thoughts
This is merely the tip of the iceberg. We cannot stop here. Over the next few weeks, I will be sharing some more thoughts interviewing for diversity and managing flexible/diverse teams.
If you like what I have to say, reach out and let me know. See you on the other side.
[^1] https://www.reuters.com/article/us-amazon-com-jobs-automation-insight-idUSKCN1MK08G [^2] https://www.youtube.com/watch?v=z9aUseqgCiY [^3] https://hbr.org/2021/09/return-to-work-programs-come-of-age