AI in HR: 5 Costly Recruiting Mistakes Companies Make (and How AI Fixes Them)
Dr. Agus Masrianto · August 23, 2026 · 5 min read
Hiring is one of the highest-stakes decisions a business makes, and yet most organizations run their recruiting process on habits formed decades ago — gut calls, keyword scans, and a résumé pile that grows faster than anyone can read it. The result is a process riddled with predictable, fixable errors.
The good news: you don't need a technical background to fix them. You need to know where the leaks are and which AI-assisted moves patch each one. Let's go through the five I see most often.
Mistake 1: Writing Job Descriptions That Attract the Wrong Candidates
Most job descriptions are copy-pasted from an old posting, loaded with jargon, and accidentally coded for a very narrow candidate profile. Phrases like "digital native," "works well in a fast-paced environment," or even unnecessary degree requirements filter out capable candidates before they ever apply.
The AI fix: Feed your draft job description into an AI tool with a prompt like this one:
"Review this job description for language that may unintentionally exclude qualified candidates based on gender, age, or educational background. Suggest neutral, inclusive alternatives and flag any requirements that may not be genuinely necessary for success in the role."
In sixty seconds you get a redline that would take a DEI consultant half a day to produce. Better input, better applicant pool — before the process even starts.
Mistake 2: Using Résumé Screening as a Proxy for Ability
Keyword-based screening — whether done by human eyes or a basic ATS — rewards candidates who know how to game the system, not necessarily candidates who can do the job. Someone who mirrors your exact job-description language scores high; a strong performer from a different industry with transferable skills gets filtered out.
The AI fix: Build a structured screening rubric before you post the role. Prompt your AI to generate a weighted scoring matrix tied to actual job competencies — things like problem-solving approach, relevant project types, or domain exposure — not just title and credential matching. Then use that rubric consistently across every résumé you read. Consistency alone removes a significant layer of arbitrary variation.
Mistake 3: Unstructured Interviews That Measure Likability, Not Fit
When every interviewer asks different questions in a different order, you are not measuring the candidate — you are measuring how comfortable that person made you feel in a room. That comfort is heavily influenced by shared background, communication style, and appearance. None of those reliably predict job performance.
The AI fix: Use AI to build a structured interview guide for each role. A prompt template to get you started:
"Create a structured behavioral interview guide for a [role title] position. Include five to seven competency-based questions, the specific competency each question is designed to assess, and a scoring rubric with behavioral anchors for a 1–5 scale."
Now every interviewer is evaluating the same dimensions with the same scale. Calibration improves. Gut-feel diminishes. Defensible hiring decisions increase.
Mistake 4: Ignoring the Candidate Experience Until It's Too Late
Most companies spend thousands of dollars sourcing candidates and then lose them during a slow, silent application process. No acknowledgment email. A two-week gap before a first call. Generic rejections with zero feedback. Candidates talk, and employer brand damage compounds quietly.
The AI fix: Map your current candidate journey from application to offer, then identify every touchpoint where communication drops. AI can draft personalized acknowledgment messages, status-update templates, and even empathetic rejection emails that leave candidates with a positive impression of your organization. The prompts take minutes to build; the brand protection lasts.
Mistake 5: Making Offers Based on What You Think You Can Get Away With
Anchoring a salary offer to a candidate's previous compensation — rather than the market rate for the role — is not only a common practice, it's now illegal in a growing number of US states, including California, New York, and Illinois. Beyond the legal exposure, it perpetuates pay gaps and poisons long-term retention.
The AI fix: Use AI to aggregate publicly available compensation data and build a transparent pay band before any offer conversation happens. A prompt like this:
"Summarize publicly available salary range data for a [role title] in [city, state] at a [company size] company, drawing from sources like the Bureau of Labor Statistics, Glassdoor, and LinkedIn Salary. Present the data as a low–mid–high band."
You walk into the offer conversation with a defensible, market-anchored number — not a negotiation anchored to someone's past.
The Bigger Picture
Each of these five mistakes shares a common root: a process designed around convenience rather than rigor. AI doesn't replace human judgment in hiring — it structures the inputs so your judgment has something reliable to work with.
None of the fixes above require you to build a model, read a line of code, or buy enterprise software. They require a clear prompt, a consistent process, and the willingness to question how things have always been done.
If you want to go deeper on all eight core AI-HR applications — including a full ethics and governance framework to keep your AI-assisted hiring legally sound and equitable — that's exactly what I cover in AIBM: AI for HR & Leadership. You'll walk away with 45 ready-to-use prompts, a structured 90-day implementation roadmap, and the confidence to lead this transformation inside your own organization.
The best time to fix your recruiting process was before your last bad hire. The second best time is now.