What Google's new AI hiring tool is secretly looking for in your resume
Google's secret AI hiring tool hasn't been as successful as the tech giant might have hoped, as a leaked internal memo suggests that it has been automatically rejecting eligible and qualified candidates, with human reviews now having to take over.
AI hiring systems have seemingly grown in popularity over the last few years as a means to filter out the often overwhelming number of applicants that any one open role receives.
There's always the fear that they aren't performing exactly how you'd expect though, with both prospective applicants and the companies looking for new employees worried that the perfect person could be accidentally passed over by a tool due to a technicality.
Those worries seemingly still exist within the heart of the AI industry itself, with Google own AI lab – Google DeepMind – appearing to admit that its automated hiring tools aren't working to the best of their ability.
Google's AI hiring tools appear to pass over perfect applicants
As reported by Bloomberg, an internal memo ironically including a disclaimer to not share the information inside widely has revealed that candidates for open roles at Google are now being encouraged to fill out a separate form specifically designed to be reviewed by a human.
Advert
The call was made by the AGI Safety and Alignment Team at DeepMind, with the disclaimer noting: "We have an applications system with a non-trivial probability your CV will be screened out incorrectly or take too long to reach us.
"Filling out this form makes sure that a real human on the team will get to see your application," the lab advises, although it does make clear that there remain no 'shortcuts' to success despite this new route.
While DeepMind has denied claims that applicants are being filtered out 'incorrectly', the disclaimer does also include advice to applicants who might be using AI to 'game the system' in reverse.
"A real human will read these," the advice warns. "These humans get really tired of reading LLM answers, because they all sound very samey."
What do AI hiring tools look for in applications?
It's impossible to know exactly what AI hiring tools like in applications as that would allow people to game the system, but it's generally accepted that keywords lifted directly from the system, alongside having the exact right experience, is going to give you a leg up when your CV is processed.
What makes it difficult for applicants, however, is that there's often little nuance when it comes to AI scanning tools, as they can completely pass over or misinterpret experience and skills that would be relevant if they are presented in a way that the model doesn't recognize.
Google offer tools similar to these for other companies through Workspace, indicating that they "save HR time by quickly creating drafts for job postings, evaluating resumes, and forecasting hiring needs," but if Google themselves are having to bypass this then it's not a good sign for the rest of the employment world.
Amid frustration and often after hundreds of rejections, candidates are also turning to AI themselves to optimize their CVs and cover letters as much as possible, as the theory suggests that nothing knows what AI is looking for more than the models themselves.
As Google's advice suggests, however, that can lead to homogeneity among applications that falls apart once viewed with human eyes — but for some, getting past that first stage of automated scanning is all that's important, with their experience hopefully standing out past that point.