Inside Google DeepMind's Own AI Hiring Glitch and the Human Workaround

Compiled by the editorial desk with reference to Bloomberg's reporting and public statements from Google DeepMind.

An internal memo from Google DeepMind, the company's flagship artificial intelligence research unit, acknowledges that its automated applicant screening system may incorrectly discard resumes, and advises job seekers to use a supplementary form to ensure their application reaches a human reviewer. The document, seen by Bloomberg, offers a rare glimpse into the practical challenges of AI-driven recruitment, even at one of the world's leading AI institutions.

The memo, which includes the unusual request "PLEASE DO NOT SHARE THIS DOC WIDELY," was circulated by DeepMind's AGI Safety and Alignment Team. It states: "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."

This admission is striking given DeepMind's role in developing cutting-edge AI technologies. The team's decision to create a workaround suggests that even those who build advanced algorithms do not fully trust automated hiring tools to identify the right candidates.

Company Response and Broader Context

In response to the memo, a Google DeepMind spokesperson told Bloomberg that the company's AI hiring software is functioning as intended. "This team set up a special form to go past the recruiter review, and get their resumes direct to the people on the team," the spokesperson said. "But there are no shortcuts to getting hired."

The incident highlights a growing concern among job seekers and HR professionals: the widespread use of AI in recruitment may inadvertently filter out qualified applicants. A 2023 survey by the Society for Human Resource Management found that nearly 40% of companies use some form of AI in hiring, yet many candidates report frustration with automated systems that fail to recognize relevant experience.

For DeepMind, the memo is particularly awkward, as it comes from the same team tasked with ensuring AI systems are safe and aligned with human values. Critics argue that if the lab's own tools cannot reliably process job applications, there are broader implications for AI deployment in other high-stakes areas.

While the memo does not specify which AI tool is used or how often errors occur, it suggests that the problem is significant enough to warrant a manual intervention. The workaround form is not part of the standard application process, meaning that only candidates who happen to see the memo would benefit—a potential equity issue in hiring.

This is not the first time Google has faced scrutiny over its hiring practices. In 2020, the company faced backlash over its AI ethics board, and in 2022, it announced a hiring slowdown amid economic uncertainty. However, the DeepMind memo is unique in that it directly implicates the company's own technology in the hiring process.

As AI becomes more embedded in recruitment, the DeepMind memo serves as a cautionary tale. It underscores the need for transparency and human oversight in automated systems, especially when they have a direct impact on people's livelihoods.