Most ATS vendors are selling you AI. Most of them are not delivering it. What they're actually delivering is Boolean search — keyword matching that has existed since the 1990s — dressed up with a new badge and a bigger price tag. If you can't tell the difference, you're paying for something that isn't doing what you think it is.
What do ATS vendors actually mean when they say "AI"?
Usually, they mean one of three things — none of which are AI.
Ranked Boolean search. The system matches keywords in a resume against keywords in a job description and generates a score. Candidate has "project management" in their resume, job requires "project management," score goes up. This is pattern matching. It has nothing to do with understanding what makes a qualified candidate.
Rules-based filtering. You set criteria — must have X years of experience, must have Y degree — and the system filters accordingly. This is a spreadsheet with an interface. It's not AI. It doesn't learn. It doesn't adapt. It does exactly what you tell it to do and nothing more.
Pre-set scoring templates. The vendor builds a scoring rubric, applies it to every candidate, and presents the output as an AI-generated match score. The score isn't generated from intelligence — it's generated from a static formula that some engineer built once and hasn't changed since.
All three of these get called AI in the marketing materials. None of them are. And the problem isn't just semantic — it's that teams make real decisions based on these scores. Candidates get filtered out. Roles go unfilled. Recruiters trust a number that doesn't mean what they think it means.
What is Boolean search and why isn't it AI?
Boolean search is a method of filtering results using logic operators: AND, OR, NOT. It has been the backbone of database search since the 1950s. It is powerful as a tool when a human is operating it intentionally. It is not intelligent. It does not understand context, intent, or nuance.
A Boolean search for "nurse AND ICU" will surface every resume with both those words. It will miss the candidate who spent six years as a critical care specialist and never used the acronym ICU in their resume. It will surface the candidate who took an ICU training course once and lists it in their certifications, even if they've never worked in a hospital.
The system doesn't know the difference. It can't. It's matching strings of text, not evaluating fit.
When one of Matt Schalsey's contacts started researching the AI market back in 2018, vendors were already pitching ranked Boolean as AI. The terminology has evolved — the underlying technology often hasn't. The badge changed. The search logic didn't.
That's the Wizard of Oz problem. There's a lot of smoke and impressive-looking controls, but there's not much behind the curtain. The recruiter trusts the AI score. The AI score is a keyword count. The best-fit candidate — the one who would have been a strong hire — got filtered at 40% because their resume didn't use the exact terminology the job description used.
What does real AI in recruiting actually look like?
Real AI in recruiting understands language, not just text. It does different things than Boolean search does — meaningfully different things that change outcomes.
Semantic matching means the system understands that "critical care nurse" and "ICU nurse" describe the same thing, even though the words don't match. It understands that a recruiter searching for "strong communicator" might be looking for someone whose resume shows leadership, cross-functional collaboration, and client-facing experience — not someone who simply wrote the phrase "strong communicator" in their objective statement.
Pattern learning means the system improves based on outcomes, not just rules. It can learn that your hiring managers tend to advance candidates with a specific combination of experience, even if that combination doesn't appear in the job description. It gets better the more it's used.
Contextual scoring means the system can weigh signals differently depending on the role, the team, the level. An entry-level candidate in food service operations shouldn't be scored against the same rubric as a senior finance analyst. Real AI adapts. A scoring template doesn't.
The difference in output is significant. When PerfectHire's Conduit evaluates a candidate, it's not counting how many times a word appears. It's reading the candidate in the context of the role and generating a score that reflects actual fit — not keyword overlap. That's a different architecture, not just a different label.
How do you tell the difference between real AI and relabeled Boolean search?
Ask five questions. The answers will tell you what you're actually buying.
- Does the system surface candidates who don't use the exact terms in the job description? If it can only find candidates who match keywords verbatim, it's Boolean.
- Does the score change based on historical hiring patterns at your company? If it's the same score formula regardless of your data, it's a template.
- Can it explain why a candidate ranked where they did in plain language — not just a percentage? If all you get is a number, there's no intelligence generating it.
- Does it get more accurate over time as your team makes hiring decisions? If the scoring never improves based on outcomes, it doesn't learn.
- Will your ATS vendor show you the underlying logic of how candidates are ranked? If the answer is vague or defensive, that's usually because the answer is "we run a keyword search."
The vendors who can answer these questions directly tend to be the ones actually doing what they claim. The ones who redirect to demo videos and case studies when you ask about the technical architecture are often the ones with the least to show.
Why does it matter if your ATS AI is actually AI?
It matters because Boolean search filters out qualified candidates — and you don't know it's happening.
Keyword-based filtering creates invisible disqualifications. A candidate is strong. Their resume doesn't use the exact phrase. The system scores them at 45%. The recruiter sees a sea of candidates, moves the ones in the 70s and 80s to the next stage, and the 45% candidate never gets looked at.
Nobody logged a bad decision. The system did what it was designed to do. The hiring manager never knew a strong candidate existed. The role eventually gets filled by someone adequate instead of someone exceptional — and no one connects the outcome back to the scoring method.
At volume, this problem compounds. High-volume environments — healthcare, hospitality, food service — send thousands of applications through the filter every month. If the filter is built on keyword matching, it's generating noise masquerading as signal. Your team is spending time on candidates the system surfaced for the wrong reasons, and missing candidates the system buried for no good reason.
The PerfectHire ATS+ was built to solve this exact problem. Not by adding an AI badge to the same old scoring model, but by replacing the scoring model with one that actually understands candidate fit. When you combine that with Conduit's semantic matching layer, the candidates that surface are the ones who are actually qualified — not just the ones who wrote their resume the way the job description was written.
That's the difference between a tool that processes applications and a platform that actually helps you hire. If you want to see what that looks like against your real open roles, book a demo and we'll show you the output side by side.
Frequently Asked Questions
How do I know if my ATS is using real AI or just Boolean search?
Ask your vendor if their candidate scoring adapts based on your company's historical hiring decisions. If the answer is no — or if they can't clearly explain how the scoring model changes over time — you're likely working with a rules-based or keyword-matching system, not genuine AI. Real AI recruiting software improves with use; Boolean-based systems don't learn at all.
What is Boolean search in recruiting?
Boolean search is a keyword-matching method that uses logic operators (AND, OR, NOT) to filter resumes against job descriptions. It surfaces candidates whose resumes contain specific words or phrases. It does not understand context, synonyms, or the underlying meaning of experience — which means qualified candidates whose resumes use different terminology are routinely filtered out.
What is semantic candidate matching?
Semantic matching is a form of AI-driven candidate evaluation that understands meaning, not just keywords. Instead of matching the word "nurse" to the word "nurse," a semantic system understands that "critical care specialist" and "ICU RN" describe similar experience. Platforms like PerfectHire Conduit use semantic matching to evaluate candidates based on actual fit rather than keyword overlap, which produces higher-quality candidate shortlists.
Why do ATS vendors call Boolean search AI?
Primarily because "AI" is a selling point and "Boolean search" is not. The underlying technology hasn't meaningfully changed for many platforms, but the marketing language has. Most buyers don't have the technical background to evaluate the difference, so vendors benefit from using AI terminology broadly. Asking vendors to explain exactly how their scoring model works — and whether it improves based on hiring outcomes — usually reveals what's actually under the hood.
Does real AI recruiting software cost more than standard ATS platforms?
The price difference matters much less than the outcome difference. A platform running Boolean search at any price is filtering out qualified candidates and generating inaccurate fit scores. A platform with real semantic matching surfaces better candidates faster, which reduces time-to-fill and cost-of-vacancy across every open role. The question isn't whether AI recruiting software costs more — it's whether the current system's filtering failures are costing you more than the upgrade would.