AI in Recruitment

Where AI is used in the hiring process

AI shows up at nearly every stage now. The main uses, roughly in funnel order:

• Job descriptions and ads: drafting postings, adjusting wording, and choosing where to spend ad budget.
• Sourcing: searching profiles and past applicants, ranking them against a role, and writing personalized outreach.
• Resume screening and matching: parsing resumes and scoring fit against requirements.
• Conversational screening: chat, SMS or voice conversations that ask screening questions and answer candidate questions around the clock.
• Interviews: an AI interview where an AI asks structured questions by voice or video, follows up, and produces a scored summary for a recruiter.
• Scheduling and logistics: booking, rescheduling and reminders.
• Assessment: scoring language proficiency, typing, simulations and work samples.
• Interview intelligence: transcribing and summarizing human interviews.

Increasingly these are bundled into an AI recruiter: software that runs several stages end to end rather than one task.

Where AI helps most

AI earns its keep where volume is high and the work is repetitive but language-heavy. A contact center receiving 6,000 applications a month for 150 seats cannot phone-screen everyone. Without AI, most applicants never speak to anyone, and the best ones accept another offer while waiting. An AI screen can talk to every applicant within minutes, check availability and language skills, and pass a ranked shortlist to recruiters the same day.

The second big gain is consistency. A structured AI screen asks every candidate the same questions and scores against the same rubric, which is more than most human phone screens manage by the fourth hour of the day.

For senior, niche or relationship-driven hiring the gains are smaller. AI can help with sourcing and notes, but closing a VP of Operations is still a human job.

Where it goes wrong

• Learning from biased history. A model trained on who you hired in the past learns what your past hiring managers preferred, including their biases. See AI bias in hiring.
• Proxies for protected traits. Zip code, school names, employment gaps and even word choice can correlate with race, sex, age or disability.
• Black-box scores. If a recruiter cannot see why a candidate scored 62, they cannot challenge it, and you cannot defend it.
• Scoring the wrong things. Tools that score facial expressions or vocal tone have weak scientific support for predicting job performance, and emotion recognition at work is now banned in the EU.
• Over-automation. Letting the AI reject people without human review removes the check that catches its mistakes.
• Candidate distrust. Undisclosed AI, especially in interviews, feels deceptive. Tell candidates when AI is involved and what it does.

How to evaluate an AI recruiting tool

Ask vendors, in writing:

1. What exactly does the model score, and what data was it trained on?
2. Can a recruiter see the evidence behind each score (transcript excerpts, answers, criteria)?
3. Has the tool had an independent bias audit, and can we see the impact ratios by sex and race/ethnicity?
4. Which decisions does it make on its own, and which require a human?
5. How are candidate data stored, retained and deleted, and is it used to train models for other customers?
6. What accommodations exist for candidates with disabilities or limited connectivity?

Then pilot it on one role, compare AI scores with recruiter judgment and later performance, and check pass rates by group before rolling out.

Laws and regulations

There is no single US federal AI hiring law, but existing law applies fully. Title VII, the ADA and the ADEA cover AI-driven selection the same way they cover any test, and the employer stays responsible for tools bought from a vendor. On top of that:

• NYC Local Law 144 (enforced since July 2023) requires an independent bias audit within the year before using an automated employment decision tool, a public summary of results, and notice to candidates.
• The Illinois AI Video Interview Act requires notice, explanation and consent before AI analyzes recorded video interviews.
• Illinois amended its Human Rights Act, effective January 2026, to bar AI use that discriminates and to require notice.
• Colorado's AI Act was repealed and replaced in May 2026 by SB 26-189. From January 1, 2027, employers whose automated tools materially influence hiring decisions must give notice, explain adverse decisions within 30 days, and offer data correction and human review.
• The EU AI Act lists recruitment and candidate evaluation systems as high-risk, with duties on risk management, human oversight and transparency.

Rules change quickly and vary by state and country. Treat this as background, not legal advice. Our article on AI in talent acquisition covers adoption in more detail.