AI in Recruitment. What’s Changing, What’s Staying, and What Can’t Be Automated

A few weeks ago, I read a post that really got me thinking. The article The Switchboard Operator Fallacy: Who will execute when AI thinks? raises a point that those of us working in Human Resources cannot ignore: AI doesn’t just replace the task; it also replaces the "escape routes" that historically existed whenever a task disappeared.

The question I ask myself from my role in recruitment is more specific, What is changing in our field, at what speed, and what is it that no system can do better than a person with experience and judgment?

What the Data Shows

We aren't looking at an emerging trend; AI in recruitment is already an operational reality for most organizations. In 2025, 43% of organizations used AI for HR tasks, compared to 26% in 2024—nearly doubling in a single year Truffle. Furthermore, 87% of companies report using AI tools in their recruitment processes, and 93% of recruiters plan to increase that usage in 2026 DemandSage.

The benefits are concrete: teams save an average of 4.5 hours per week by automating repetitive tasks (Shortlistd), and companies report a 30% to 50% reduction in time-to-hire (Hirebee). At first glance, these numbers sound like good news for everyone.

The Problem No One Wants to Mention

When something moves this fast, unintended consequences quickly follow. 72% of organizations face difficulties with transparency requirements regarding AI use in selection, and 66% struggle to create diverse datasets to train models equitably Second Talent.

In practical terms, a system trained on historical data can reproduce and amplify the biases of those who hired in the past. 35% of recruiters fear that AI might exclude candidates with unique skills or unconventional career paths DemandSage. When we automate the initial filter without human judgment, we risk discarding the exact profile that would have made the difference.

Where Human Judgment is Irreplaceable

AI can process thousands of CVs per hour, detect patterns, and identify matches between profiles and requirements. All of that is real and welcome. However, there are dimensions of the selection process that no algorithm can evaluate accurately.

The real motivation behind an application isn't on the CV. Cultural fit requires human nuance, conversation, and trained intuition. Undemonstrated potential, growth capacity, and attitude toward a challenge are part of the recruiter’s craft, not a formula. Furthermore, the way a company treats its candidates says everything about how it treats its employees. A chatbot cannot convey that message.

The key takeaway from the 2025 AI in Recruitment Report by Insight Global is clear, while AI improves the process by streamlining tasks and providing valuable insights, it is the human recruiter's experience that ensures the technology is used effectively Insight Global.

Our Vision, Humans at the Center of the Process

At Converis, we have a clear stance on this, technology should serve judgment, not replace it. We call this working with the "human-in-the-loop", ensuring that at every critical stage of the process, there is a person responsible for interpreting, deciding, and caring.

We use tools that speed up the search and the initial filter. But the heart of our process—the conversation with the candidate, reading the client’s context, and the final recommendation—is profoundly human. Always.

Returning to the question, who executes when AI thinks? In talent selection, the answer has important nuances. AI can process and classify, but the decision of which specific person will join a specific team still requires trained, contextualized judgment with a history behind it. Roles that require building trust, reading context, and making decisions that impact real lives are not disappearing. They are transforming. At Converis, we work every day to meet the challenge of that transformation.

By 2030, it is projected that 94% of recruitment processes will incorporate AI at some level (Second Talent). That doesn’t mean a machine will make 94% of the decisions. It means those of us working in talent must learn to use these tools with intelligence, ethical judgment, and an awareness of their limits. That is what we are building. And that is what is worth defending.