Overview

Lecture URL: https://skill.re/learn/recruiting/verification-techniques-spot-checking-facts-sources-and-candidates.php

TRANSCRIPT: Verification Techniques: Spot-Checking Facts, Sources, and Candidates

Course: AI for Recruiters - Professional Credential

Module: Level 2: Hands-On Foundations

Section: Chapter 9 -- Reviewing AI Output Critically

Theme: reviewing-ai-output-critically

Lecture: 9.3

Duration: 60 min

Format: Workshop + Hands-On

Audience: Recruiters beginning to use AI tools

Prerequisites: L1 Certification

What you will learn: You'll develop systematic approaches to verify AI outputs: fact-checking

claims, validating sourcing, confirming candidate details. By the end, you'll have verification

frameworks that prevent errors from influencing hiring decisions.

INTRODUCTION

Verification is the systematic work of checking whether what AI told you is actually true.

This is different from spot-checking (looking for obvious errors). Verification is deliberate,

methodical, and based on source material. When AI claims something, you verify it against: the

resume, the notes, the LinkedIn profile, the reference, the actual reality.

This takes time. It's worth it because it prevents bad decisions based on false information.

Today, we're building your verification toolkit.

THREE VERIFICATION DOMAINS

You need to verify in three domains: facts about candidates, sources used in research, and claims

made in assessments.

DOMAIN 1: FACTS ABOUT THE CANDIDATE

These include: years of experience, specific technologies, specific projects, locations, education.

Verification process:

1. Identify every specific fact in AI's output
2. Check against: resume, interview notes, LinkedIn profile
3. Flag any discrepancies
4. Resolve through follow-up if needed

Example:

AI says: "8 years of backend engineering experience"

Resume says: "5 years as engineer, 2 years as tech lead"

Resolution: Verify with candidate. Is it 5 or 7 years of backend specifically?

DOMAIN 2: SOURCES AND RESEARCH

When AI is researching candidates (sourcing, skill assessment), verify the sources.

Verification process:

1. Check: Does the source actually exist?
2. Check: Is the information attributed correctly?
3. Check: Is the interpretation accurate?

Example:

AI says: "Active on GitHub with 50 repositories"

Verify: Actually check the GitHub profile. Is it 50 repos? Are they relevant?

DOMAIN 3: CLAIMS AND ASSESSMENTS

When AI makes claims about capability, fit, or concerns, verify the evidence.

Verification process:

1. Identify the claim ("Strong problem-solver")
2. Find the evidence cited
3. Check: Does the evidence support the claim?
4. Alternative explanation: Could the same evidence mean something else?

Example:

Claim: "Strong problem-solver"

Evidence cited: "Debugged a complex system issue"

Verification: Does solving one problem mean strong problem-solving generally? What else would you

need to verify?

SPOT-CHECKING FRAMEWORK

You don't have time to verify everything. Develop a spot-checking framework:

TIER 1: ALWAYS VERIFY

• Specific quantitative facts (years of experience, team size, scope)
  - Major qualifications (languages, frameworks, key skills)
  - Facts that would change your decision if wrong

TIER 2: OFTEN VERIFY

• Assessments of soft skills
  - Claims about motivation or fit
  - Sourcing accuracy for passive candidates

TIER 3: SPOT CHECK

• General descriptions
  - Background narrative
  - Non-critical details

Use this framework to allocate your verification effort where it matters most.

VERIFICATION TOOLS AND TECHNIQUES

TECHNIQUE 1: RESUME VERIFICATION

Pull the resume. Compare AI output facts against resume facts. Document discrepancies.

TECHNIQUE 2: INTERVIEW NOTES VERIFICATION

Pull the notes. For every claim about what the candidate said or did, verify against actual notes.

TECHNIQUE 3: INDEPENDENT SOURCE CHECK

For sourcing research, actually verify the claimed sources. Is the GitHub real? Does it match the

description? Is the open source contribution actually attributed correctly?

TECHNIQUE 4: REFERENCE VERIFICATION

For senior roles, consider brief pre-call reference conversation to verify key claims: years of

experience, scope of projects, specific skills.

TECHNIQUE 5: FOLLOW-UP CONVERSATION

If AI makes claims that matter and you can't verify them, ask the candidate directly: "Can you

walk me through your experience with X?"

ANTI-PATTERNS

ANTI-PATTERN 1: NOT VERIFYING CRITICAL FACTS

Description: Accepting AI's facts without checking, especially when they're specific numbers.

Why it fails: Specific facts are most likely to be hallucinated. If you don't verify, you're

basing decisions on false information.

How to avoid: Always verify Tier 1 facts.

ANTI-PATTERN 2: ASSUMING SOURCING IS ACCURATE

Description: Using AI's description of a candidate found through research without verifying the

source.

Why it fails: AI might misattribute or misinterpret what it found.

How to avoid: Independently verify sourcing. Actually look at the GitHub, the blog post, the open

source profile, whatever was claimed.

ANTI-PATTERN 3: NOT ACCOUNTING FOR CHANGED INFORMATION

Description: Using AI output from research done weeks ago, not accounting for the fact that things

might have changed.

Why it fails: Candidate's background or GitHub might have been updated. You're using stale

information.

How to avoid: For high-priority candidates, re-verify information close to decision time.

PRACTICE PROMPTS

Exercise 1: Build Your Verification Checklist

For a role you're hiring for, list: (1) Tier 1 facts (always verify), (2) Tier 2 claims (often

verify), (3) Tier 3 details (spot check).

Exercise 2: Verify an AI Output

Take an AI summary or assessment. Go through Tier 1 facts. Verify each against source material

(resume, notes, etc.). Document discrepancies.

Exercise 3: Source Verification Practice

AI claims research about a candidate ("Active in open source," "Shipped 5 products"). Independently

verify each claim. What's accurate? What's inaccurate?

Exercise 4: Develop Your Verification Protocol

For your team's hiring process, design a verification protocol: What gets verified? When? By whom?

How much time does it take?

Exercise 5: Create Your Verification Template

Build a template you can use to systematically verify AI outputs. What information goes in it? What

decisions does it support?

KEY TAKEAWAYS

1. Verification is systematic fact-checking of AI outputs against source material.
2. Verify in three domains: facts about the candidate, sources used in research, claims made in

assessments.

1. Don't verify everything. Use a tiered approach: always verify critical facts, often verify

important claims, spot-check everything else.

1. Specific quantitative facts are most likely to be hallucinated. Always verify years of

experience, team size, scope.

1. Independently verify sourcing. Actually check the GitHub, the blog post, whatever was claimed.
2. For high-priority candidates, re-verify information before final decisions.

GLOSSARY

Verification: The systematic process of checking whether AI-generated information is accurate by

comparing against source material.

Spot-Checking: Sampling-based verification. You don't verify everything, but you check key facts.

Tier 1 Facts: Critical facts that would change your decision if wrong. Always verify.

Tier 2 Claims: Important but not critical. Often verify.

Tier 3 Details: Non-critical information. Spot-check only.

SYNTHESIS AND APPLICATION

Verification is your safety mechanism. When you systematically verify AI outputs, you catch errors

before they influence hiring. It takes time, but it's the difference between AI-assisted hiring and

AI-misled hiring.

This week, implement a verification process for one role you're hiring for. Choose a candidate.

Verify Tier 1 facts. See how long it takes. See what you find.

REFLECTION EXERCISE

1. What AI-generated information do you currently trust without verifying? Should you be verifying

it?

1. For your recent hiring, if you had verified AI outputs, what might you have caught?
2. What's the time cost of verification? Is it worth it for your hiring?
3. How might you build verification into your team's process, not just as a personal practice?

CLOSING REMARKS

Good hiring is built on verified information. AI is a tool, but you're responsible for the accuracy

of what goes into your decisions. In the final chapters, we're moving from reviewing AI outputs to

building systems and cultures that support responsible AI use.

AI for Recruiters Certification Program

Level 2: Hands-On Foundations | Reviewing AI Output Critically | Lecture 9.3

A SkillsClinic initiative.

Duration: ~60 minutes | Word Count: ~2,380