Even at Google, there are fears that AI will reject the wrong candidate
Google DeepMind researchers have acknowledged that internal AI systems may mistakenly reject job applications. As a result, they are asking candidates to fill out a special form to ensure their resumes reach human recruiters.
Although Google promotes its AI tools to organizations as a way to help them quickly review all submitted resumes and find the most promising candidate for every role, its own AI researchers have cast doubt on the reliability of these tools for recruiting, according to a report by Bloomberg. The AGI Safety and Alignment team at Google DeepMind, which works to reduce the risks associated with advanced artificial intelligence, encourages candidates for open positions on the team to fill out a special form alongside their application to prevent the possibility that their candidacy will be automatically filtered out by the company's internal AI systems, according to a document obtained by Bloomberg, which includes the caveat: "Please do not share this document widely."
"We have an application system where there is a non-negligible probability that your resume will be rejected by mistake or that it will take too long to reach us. Filling out this form ensures that a real person on the team will see your application," the document states.
However, they warn candidates against using AI themselves: "That same person is tired of reading answers from AI models, because they all sound the same."
A spokesperson for Google DeepMind rejected the claim that the company's systems inaccurately filter candidates:
"The team created the special form so that it would bypass the screening process, and so that resumes would be sent directly to team members. But there are no shortcuts to getting hired," he emphasized.
While more and more companies rely on AI for screening and recruiting, criticism of these systems is growing. Another concern is that they discriminate against candidates. For instance, a check conducted by Bloomberg found that OpenAI's ChatGPT showed signs of potential discrimination based on candidates' names. Similarly, a lawsuit was filed against Workday Inc alleging that its placement system filters candidates based on race and age, in violation of the law. The company denied the allegations and claimed that ultimately, it is humans who make the decisions about hiring. At the same time, some job seekers have already learned methods to use AI to their advantage by manipulating the filters.
