Explainer

How to Spot Low-Substance Resumes

ChatGPT-written resumes are flooding recruiter inboxes. Here's what to actually look for, and how Tersen flags them automatically without accusing anyone of anything.

By Tersen Team5 min readUpdated Nov 2026

The Problem

Generative AI has made it trivially easy to produce a resume that reads well but says almost nothing specific, and recruiters are seeing more of them every month.

A resume generated (or heavily "polished") by an LLM often looks professional at a glance: clean formatting, confident phrasing, no typos. But strip away the polish and there's frequently nothing underneath: no specific project, no number, no tool named, no outcome described. It's the resume equivalent of a form letter.

That's a real cost for a recruiter screening at volume: time spent reading a well-formatted resume that ultimately tells you nothing you can evaluate, verify, or ask about in an interview.

Warning Signs to Look For

None of these alone makes a resume low-substance, but several appearing together is a strong signal the resume is generic, whatever its origin.

Generic buzzwords, no specifics

"Results-driven team player with strong communication skills" appears everywhere and describes no one in particular.

Vague responsibilities

"Responsible for managing projects" instead of what project, what scope, what outcome.

Zero quantified achievements

No numbers anywhere: no team size, no percentage improvement, no dollar figure, no timeline.

Suspiciously uniform structure

Every bullet is the same length, same sentence shape, same rhythm, a pattern large language models default to.

Buzzword density without evidence

"Leveraged synergies to drive innovation" reads well out loud but describes nothing you could verify in an interview.

Mismatched specificity

The job titles and dates are precise, but the actual work described stays deliberately generic.

How Tersen Flags It

Tersen runs a deterministic specificity check on every resume, looking for the exact signals above: buzzword density, vague-responsibility phrasing, and the presence (or absence) of quantified achievements. Resumes with a low specificity score and vague phrasing get a "Low substance" flag, visible right in your candidate list.

A single toggle, "Hide low-substance resumes", lets you hide flagged candidates from your view entirely while you triage, so you spend your limited attention on resumes with something concrete to evaluate.

We're deliberately honest about what this is. It's a pattern-based signal on resume text, not a literal "written by AI" classifier, since no tool can reliably prove authorship. What it reliably measures is specificity: does this resume give you something you can actually evaluate and ask about, or is it just well-formatted filler?

Nothing is ever auto-rejected. The flag surfaces in your filters and on the candidate card, and you always decide whether to look closer, deprioritize, or ignore it entirely.

See it flag a resume live

Try the interactive demo. No sign-up required.

Common Concerns, Addressed

Does this unfairly flag non-native English speakers?

The signal looks at specificity and evidence: concrete numbers, named tools, described outcomes, not grammar, vocabulary, or writing style. A resume can be flagged for being generic in perfect English, and a resume with imperfect English but real, specific achievements won't be flagged. It's not a language-quality check.

Is this accusing candidates of cheating?

No. It's a specificity signal, not an authorship claim. We deliberately don't call it an "AI detector" for this reason: literal AI-authorship detection is unreliable and unverifiable. What we can measure directly is whether a resume relies on generic phrasing instead of concrete, checkable evidence, which is useful to know regardless of how it was written.

How does the specificity check actually work?

A deterministic scoring pass reads each resume for buzzword density, vague-responsibility phrasing, and the presence (or absence) of quantified achievements, and produces a specificity score. It's a pattern-based heuristic on resume text, the same kind of signal a sharp recruiter would notice reading manually, just applied consistently across every resume in your pipeline.

Can it be wrong?

Yes. Like any heuristic, it can miss a genuinely specific resume written in a terse style, or flag a resume that's generic for reasons other than AI generation (a rushed application, a template the candidate didn't customize). That's why it's a filter you control, not an automatic rejection: you see the flag, and you decide whether to look closer or skip it.

Can ATS systems or recruiters detect AI-generated resumes?

Not reliably. Some tools claim to, but AI-writing detectors are known to misfire, often on people writing in a second language. A polished resume isn't a bad candidate either. The useful question is whether the claims are specific and backed by evidence: numbers, named tools and real outcomes. That's what Tersen's specificity score checks, whoever wrote the resume.

The Bottom Line

You don't need to solve "did AI write this" to solve the real problem, which is: does this resume give me evidence I can evaluate? A specificity signal answers that question directly, consistently, and without making an accusation you can't back up.

Combine that with evidence-based scoring against your Role Card, and you spend your screening time on resumes worth reading, not on filler.

Related Reading

See It in Action

Try our interactive demo to see the specificity score and evidence-based scoring together. No sign-up required.