Navigating AI and the future of talent

AI is rapidly changing the future of work. Like many of us, I have witnessed its impact on my workflow. As MVL’s Head of Talent, I use AI to automate mundane tasks, create more comprehensive and inclusive job descriptions, and personalize email outreach messages. It analyzes interview feedback and prepares me for onsite debrief discussions. I've been awestruck by its impact on my productivity, and I know many of the current capabilities are just the beginning; AI’s potential to further reduce burdensome administrative tasks and increase organization excites me. Through technological advances, I hope we can enhance hiring efficiency, improve candidate decision-making, and drive better business and personal outcomes in the hiring process.
‍
But while AI offers numerous benefits for recruiting, engaging in a balanced discussion about its potential risks is crucial. In the startup ecosystem, we must not overlook some of the drawbacks of AI’s application to talent acquisition, including the loss of personal judgment, bias amplification, detraction from the candidate experience, and security and privacy. By maintaining a critical approach, we can navigate these challenges and ensure the best outcomes for our teams and candidates. I’ll delve into some of my chief concerns below.

Loss of human judgment

Recruiting is inherently a human-centric process, where understanding the nuances of a candidate’s experience is crucial. While AI can analyze data and identify patterns, it cannot make the nuanced judgments experienced recruiters can about translatable experiences. A candidate who may not have the exact skills listed in the job description could still excel in the role due to a less obvious qualification. For example, when I worked at an executive search firm, many of our company’s top performers came from the hospitality industry. While they had no prior exposure to talent acquisition, they demonstrated high adaptability, perseverance, and customer-centricity. An AI system focused solely on matching keywords and preset experiences might dismiss those people altogether.

In addition to lacking the understanding of skill-based nuance, AI is also limited in its ability to make characterological judgments about candidates. AI processes large datasets and identifies surface-level patterns effectively; however, it struggles to assess traits like emotional intelligence, leadership potential, or how well a candidate might mesh with a team on a personal level. A candidate’s perseverance, ethical grounding, or collaborative spirit may be entirely overlooked by an algorithm focused purely on technical qualifications and experience. These qualities, often critical in determining long-term success within a company, require an understanding of human behavior and intuition that AI simply cannot replicate. 

Amplifying bias

As described above, AI excels at patterning and processing. I worry that over-reliance on these strengths could lead to a homogenous workforce where only those who fit a narrow set of criteria are considered. AI systems are only as good as the data they are trained on, and if that data contains biases, the AI will replicate and exacerbate those trends in decision-making.

Suppose an AI system is trained on historical hiring data from a company that has predominantly hired a particular demographic. In that case, it may learn to favor candidates who fit that demographic, thereby perpetuating discrimination. Consider the cautionary example of Amazon, a leading giant in the industry. Their innovative AI-driven hiring system hit a significant obstacle: it favored male candidates for technical positions. This bias stemmed from long-standing gender disparities within the company and the tech industry. The algorithm, trained on human-generated data, unintentionally learned these existing biases and continued reinforcing them in its hiring recommendations.

Some argue that AI could enable us to make more informed decisions without factoring in a candidate's name, gender, or educational background. AI indeed has the potential to remove certain overt biases by anonymizing aspects of a candidate’s profile—such as name, gender, or educational background—which might unconsciously influence human recruiters. In theory, it could help us level the playing field. However, this assumes that the training data is neutral and representative; unfortunately, this is rarely true. Even after stripping away demographic details, AI may still pick up on proxies for race, gender, or socioeconomic status—gaps in employment or the types of previous employers—thus reinforcing existing biases rather than eliminating them. 

Yes, human recruiters undoubtedly perpetuate systematic societal biases. Sometimes, these behaviors and beliefs are conscious and sometimes unconscious. The reason AI bias concerns me more than human bias is the scale of its decision-making power. Additionally, when we place responsibility on AI, we can be lulled into a false sense of security that we have “solved” discriminatory hiring practices by outsourcing them to an “unbiased” machine. The danger lies in believing that technology alone can eliminate bias when it can silently embed and amplify existing prejudices on a far larger scale than any individual recruiter ever could.

Candidate experience

A positive candidate experience is critical in attracting and retaining top talent because it bolsters a company's reputation in the market. There are certainly times and places for candidate-facing technology; AI-powered chatbots and automated communication systems can help hiring teams stay organized, quickly synthesize the results of skill-based assessments, and provide timely follow-up to candidates. 
‍
But, these tools often lack the empathy and understanding human recruiters can offer. Too frequently, using AI in recruiting can lead to a depersonalized and frustrating experience for candidates. AI systems that rely on standardized assessments or rigid criteria may fail to recognize individual candidates' unique strengths and potential. This can result in a poor candidate experience, where highly qualified individuals are rejected based on arbitrary factors. In a world where employer branding is increasingly important, companies that rely too heavily on AI in their recruiting processes risk damaging their reputation and losing top talent.

Even with AI’s benefits in prompt follow-up and efficiency, many candidates only appreciate the upsides of the technology after interacting with a human and being recruited in a personalized, high-touch way. This is especially true for passive, hard-to-recruit talent pools. These people receive countless daily outreach messages,—including emails, LinkedIn InMails, texts, and phone calls. They especially value having what John Vlastelica calls a “live career conversation” with “a real person who talks to me, listens to me, makes me feel like a human, and credibly answers questions I can use to make a great decision about this major life choice” (Check out John’s piece, “Will AI usher in an era of inefficiency in recruiting?” for more of his excellent thought leadership.) 

John’s assessment resonates with my own experience. As a recruiter, I often hear about confidential details of an individual’s personal life: I was the first to learn of someone’s pregnancy outside of their nuclear family; I sat consoling one man for almost an hour who cried about his frustration with the American immigration system; I was horrified by one executive's pattern of claiming women and people of color’s work on his team as his own as told to me by a former direct report who was terrified at the prospect of collaborating with her previous boss again. 

Would some people feel comfortable confiding these details to a ChatBot? Maybe. But candidates share their stories with me because I make space to engage with people and learn what matters to them. Rather than treating candidates like a means to an end, I recognize their humanity. When we over-index on efficiency, I worry we lose this deeply personal connection that is the foundation of any successful recruiting process. I worry we lose trust.

Security & privacy

This erosion of trust is not limited to the candidate-recruiter relationship but extends to how personal information is handled in the recruiting process. Security and privacy concerns become paramount as AI systems process and store vast amounts of sensitive data, including resumes, employment history, and social media profiles. This data is often stored in centralized databases, making it a prime target for cyberattacks. In a data breach, sensitive information about candidates could be exposed, leading to legal and financial repercussions for the company.

Even disregarding the potential for security breaches, using AI to analyze personal data raises ethical questions about privacy and consent. Candidates may not be fully aware of how their data is being used to train an employer’s AI-driven hiring system. The company could use information they shared with the expectation of confidentiality for purposes far beyond their intention or consent.

Conclusion

To reiterate, I’m not against the adoption of AI. I am optimistic about technology's potential to improve the future of work, including in my industry. I see three promising initial wedges for AI in recruiting: organization of talent workflows, streamlining candidate communication and scheduling, and interview loop creation and synthesis of feedback. These applications enhance efficiency and augment human capabilities without relinquishing control over critical decision-making.
‍
Outside these domains, we should proceed with cautious AI adoption in talent acquisition and not lose sight of its potential negative impacts, especially regarding the human-based processes and decisions inherent in team building. Most importantly, we must ensure that technology supports, rather than undermines, the personal connection and the trust at the heart of recruiting. 

To learn more about MVL, read our manifesto and let’s build a company together.