The Mirage of Efficiency: Has Artificial Intelligence Improved Recruitment?

I have been working in technical recruiting since 2008, and I can assert something that contradicts prevailing common sense: it has never been easier to generate candidates, and yet, hiring has never been worse. Artificial intelligence promised to free companies from the tedious part of the selection process: filtering resumes, drafting the first contact, organizing the funnel of applicants. At that point, it delivered. What no promise included is that, by reducing the volume to make it almost free, the technology ended up making criteria irrelevant. And without criteria, recruiting ceases to be a decision and becomes industrial-scale noise.

Previously, contacting forty candidates for a senior search took a week. This friction forced one to think about whom to write to and why. Today, a tool sends those same forty messages in forty seconds, personalized with a first name and a line extracted from LinkedIn. The cost of contacting the wrong person has dropped to zero, and what costs nothing is done without prior reflection.

The candidate also automates. An engineer told Wired that he sent five thousand applications with a bot and got twenty interviews. A 0.5% effectiveness rate. Manually, applying to two hundred or three hundred, he achieved those same twenty. Ten percent. The volume did not improve anything; it only hid the deterioration behind a larger number.

In the industry, they have already named it: the "automation loop." Human resources departments incorporate AI filters because they receive applications made with AI, and those filters push more candidates to automate to evade them, and the wheel turns on its own.

It is important to be precise about what this technology can and cannot do. Artificial intelligence effectively resolves the first half of the process, how to identify who masters a certain programming language or has worked in a specific industry. What it does not resolve is the second half, which actually defines a hiring: judgment. No model can anticipate whether a person will tolerate the ambiguity of a growing company, whether they will clash with a founder prone to micromanagement, or whether frequent job changes are a red flag or the least relevant data in their trajectory. And when the filter runs over these systems, even those who audit them cannot explain why a candidate was discarded. Judging still requires a conversation, a hypothesis, and the willingness to make mistakes and recalibrate.

A survey from Harvard Business Review showed that 91% of leaders consider it critical to hire well, but only 28% believe their organization does it well. That gap did not originate with AI. Artificial intelligence simply deepened it by installing the illusion of doing more when doing the same, worse, and faster. With all this speed, filling a position takes longer than ever. The average has already reached a maximum of forty-four days, and a third of hires still come through referrals, the most artisanal channel, the one no bot has moved.

I see it every week in companies that arrive exhausted at the end of a search process. They contacted three hundred people, scheduled twenty interviews, and still have not closed the search. They lack activity; they lack a diagnosis. No one defined what problem that position solves or what signals disqualify someone even if the CV shines. Without that, AI only accelerates confusion.

With automation, many companies have discontinued investment in junior profiles. Entry-level job postings continue to be published, but actual hires in that segment have collapsed because it is cheaper to ask a model for ten searches than to spend time training someone without experience. The consequence is the silent erosion of the pipeline of senior professionals for the next decade.

None of these observations imply a rejection of technology. The central point is different. Artificial intelligence amplifies what the organization already was before adopting it. If the selection process was well designed, the tool makes it faster. If it was not, it allows for industrial-scale escalation of the previous lack of criteria.

Recruiting well has never been a volume problem. It has always been, and continues to be, a precision problem. In our consulting firm, we have a rule: if out of every four candidates we present, at least one does not advance, we stop and recalibrate the brief before adding names. That pause is precisely what AI invites us to skip, and it is what should not be overlooked. The question organizations should ask themselves is not how many candidates they contacted this week, but how many they knew how to choose correctly last year. That answer, still, no model generates.

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.