This Portland startup is using AI to root out bias from recruitment

Recruitment bias in AI first came to the spotlight in 2018, when it was revealed that an Amazon program discriminated against female candidates. While it would be another 5 years before generative AI become mainstream, surfacing with it a whole new world of biased discrimination, many companies have since been working on leveraging machine learning models to find the best hires. After all, the problem that Amazon was attempting to tackle has not gone away: how to find the best talent from pools that might contain thousands of candidates?

HirelyAI is one such company. As a matter of fact, in our interview with one of its Co-Founders at CES (below, edited for clarity), he told us that ensuring AI-driven recruitment is not biased was very much front and centre to the company’s mission.

If you’ve read our piece on what 200 images of people at work reveal about bias in AI, then you’re unlikely to be surprised that we find that music to our ears. And while HirelyAI might be in its infancy, it’s a startup that we’ll be keeping an eye on.

Could you please introduce yourself and the company?

I’m Andrew Hallberg, Co-Founder of HirelyAI, a startup from Portland, Oregon that I started with Co-Founder Malik Kahel. We like to say that we’re building an AI recruiter but really we’re focused on that very first interview. That initial qualification stage can be the hardest part for both the company hiring and the applicant.

How does HirelyAI go about that?

If you’ve got a job opening you might be getting 200 or 2000 applications. A good recruiter might be able to look at 100 resumes of which, maybe, they’ll call 3 and get one of those candidates to convert.

We allow clients to create an AI interview experience. From there, we send links to all 2000 candidates, and then they can all do an initial interview. We then pull a bunch of data from those interviews, the resumes, and custom questions and do statistical analysis with the goal of providing the clients with a more diverse list. At the same time, on the backside of that, we can provide feedback to the 90% of folks who are not getting callbacks, helping them do better next time and highlighting areas where they have gaps.

You mentioned that you guys send a link to the candidates – what do they get? What is the experience like for them?

Upon receiving a link and a code, candidates can log in and then depending on what the hiring manager wants, they can set up custom questions, video questions or a chat experience. It’s meant to help sort through candidates that are serious and candidates that aren’t.

There’s been a lot of talk in the industry about recruitment being biased. What are you doing about it?

It’s actually one of the core reasons that got us started. We are trying to remove bias from the hiring process. As it is, we’ll openly say that hiring is biased. You will pick people that are similar to you. You might see someone who worked at a company you worked at or went to your same school and that leads to an interview.

The way that we’ve been combating it is, on the data side, we are stripping out a lot of data that might lead to bias and not feeding it to the AI. That’s data like gender, age or race for example. Our goal is to let the AI focus on skills assessment.

We are still working on it. We are a startup and this is something that is part of our mission. We are seeing encouraging progress. We think ultimately bias can be removed from the recruitment process, but as developers, we have to be aware of it to tackle it.

You mentioned you’re a startup – how long have you been in business? And how’s it going so far?

We got started about 3 weeks before ChatGPT went mainstream, so about a year. It’s been going really well, we’re getting a lot of traction and feel lucky to be here at CES. We’ve got quite a bit of support from different organisations and are working with a couple of partners already. We’re looking for more partnership opportunities, for ways to expand and to talk to people who are interested in what we’re doing and want to see our demos.

Recommended reading: Hiring for success: the case for skills and attitude