A bunch of people forwarded me the New York Times article about the lawsuit against Eightfold AI. In case you didn’t read it, the claim is that the AI-driven hiring software discriminated against older workers by screening them out before any human ever saw their applications.
The essence is that someone with years of experience wouldn’t stand a chance once the algorithm decided they didn’t fit the pattern. The hiring managers just saw a ranked list of candidates and trusted that the AI surfaced the “best” ones.
This is exactly the kind of thing that keeps me up at night. Not just because I run a hiring software company, but because it’s solving the wrong problem in the worst possible way.
Automate the right problems
When I talk with franchisors about AI in hiring, I always ask: What’s the actual problem you’re trying to solve?
Usually it’s that franchisees are drowning in the administrative work of hiring like getting job postings live, screening for basic requirements like licenses or certifications, the endless back-and-forth emails to schedule interviews.
All of that can and should be automated.
But evaluating whether someone should be hired? That isn’t something you want to hand off to an algorithm.
Dallas vs. Boston
We’ve worked with thousands of franchise locations over the years and developed some best practices that help ensure consistency across a system. Prescreen questions to identify whether candidates meet the minimum requirement, for example.
But there are nuances only a human can discern. The profile of the best hire in Dallas is different from the best hire in Boston. Different markets, different challenges, different team dynamics. An algorithm trained on patterns can’t evaluate that context.
It can flag keywords on an application, but it can’t understand whether this specific person can handle what this specific location needs right now.
Interviews are conversations
An interview is where you figure out if someone can do the job, will do the job, and how they’ll fit with your team. You’re asking follow-up questions, listening for specific examples, imagining how they’d interact with your customers, gauging whether they actually want to work for you or just need any job.
Even if a franchise brand takes advantage of one-way audio interviews to expedite the phone screening process — which allows managers to easily review 20+ candidates in one go — those managers are still listening to the responses to decide who to move forward to an actual interview.
You can’t outsource that kind discernment that to a bot.
Some companies are experimenting with video interviews where candidates record answers and AI then analyzes their responses to predict job performance. For frontline workers especially, this absolutely doesn’t work.
Great hires don’t come solely from algorithmic pattern matching. They come from a real human having a conversation with another real human.
More soon,
Clint