AI for HR & Recruiters
Module 5: Screening and Candidate Evaluation
Module Overview
Resume screening is one of the most time-consuming tasks in recruiting. With AI, you can streamline this process while maintaining quality and reducing bias. This module covers how to use AI responsibly for candidate evaluation.
Learning Objectives:
By the end of this module, you will be able to:
- Use AI to assist with resume screening
- Create objective evaluation criteria
- Compare candidates fairly
- Identify and mitigate screening bias
- Build efficient screening workflows
Estimated Time: 45-60 minutes
5.1 The Screening Challenge
The Volume Problem
Consider these statistics:
- Average job posting receives 250+ applications
- Recruiters spend 6-7 seconds on initial resume review
- 75% of resumes are rejected before reaching hiring managers
Traditional Screening Issues
Speed vs. Quality:
- Fast screening misses qualified candidates
- Thorough screening takes too much time
- Inconsistent criteria across reviewers
Bias Risks:
- Name bias
- School/company prestige bias
- Employment gap bias
- Format/presentation bias
Where AI Can Help
AI excels at:
- Extracting structured information from resumes
- Consistent application of criteria
- Identifying keyword matches
- Summarizing qualifications
AI struggles with:
- Understanding nuanced career transitions
- Evaluating potential over credentials
- Recognizing non-traditional experience
- Making final hiring decisions
5.2 Creating Evaluation Criteria
Before You Screen
Define clear criteria before looking at any resumes:
Essential Requirements: What absolutely cannot be missing?
- Specific certifications
- Legal requirements
- Non-negotiable skills
Important Qualifications: Strong preferences that influence ranking:
- Years of relevant experience
- Industry background
- Technical skills
- Education level
Nice to Have: Differentiators between similar candidates:
- Additional certifications
- Specific tool experience
- Language skills
AI Prompt for Criteria Development
Help me create screening criteria for a [Job Title] position.
Role requirements from job description:
[Paste key requirements]
For each requirement, help me:
1. Define what "meets requirement" looks like
2. Identify how to verify it from a resume
3. Distinguish between essential, important, and nice-to-have
4. Create a scoring rubric (1-5 scale)
Also flag any requirements that might introduce
bias and suggest alternatives.
Scoring Rubric Example
Create a resume scoring rubric for [Job Title].
Categories to evaluate:
1. Relevant experience (weight: 30%)
2. Technical skills (weight: 25%)
3. Industry knowledge (weight: 20%)
4. Education/certifications (weight: 15%)
5. Career progression (weight: 10%)
For each category, define:
- 5 points: Exceptional
- 4 points: Exceeds requirements
- 3 points: Meets requirements
- 2 points: Below requirements
- 1 point: Does not meet requirements
Include specific examples for each score level.
5.3 AI-Assisted Resume Analysis
What AI Can Do Safely
Information Extraction:
Extract the following information from this resume:
- Years of total experience
- Years of relevant experience in [field]
- Technical skills mentioned
- Highest education level
- Most recent job title and company
- Key achievements with metrics
Format as structured data I can use for comparison.
[Paste resume content]
Requirement Matching:
Compare this resume against our job requirements.
Requirements:
1. 5+ years of product management experience
2. Experience with B2B SaaS products
3. Track record of launching products
4. Strong analytical skills
5. Experience working with engineering teams
For each requirement, indicate:
- Met / Partially Met / Not Met
- Evidence from resume (quote specific text)
- Any questions to clarify in interview
[Paste resume content]
Summarization:
Summarize this candidate's qualifications in 3-4 sentences.
Focus on:
- Most relevant experience for [role]
- Key strengths
- Potential concerns or gaps
Do not make recommendations—just summarize objectively.
[Paste resume content]
What AI Should NOT Do
Never let AI:
- Make final accept/reject decisions
- Score candidates without human review
- Access personal demographic information
- Replace human judgment on potential
Always ensure:
- Human reviews all AI recommendations
- Criteria are applied consistently
- Bias is actively monitored
- Borderline candidates get human attention
5.4 Avoiding Screening Bias
Types of Resume Bias
Name Bias: Studies show resumes with "white-sounding" names get more callbacks.
AI Risk: AI trained on biased data may perpetuate this. Mitigation: Use blind resume review when possible.
Affinity Bias: Favoring candidates from the same schools or companies.
AI Risk: AI may learn to prefer prestigious institutions. Mitigation: Focus on skills and achievements, not pedigree.
Gap Bias: Penalizing employment gaps.
AI Risk: AI may flag gaps as negative. Mitigation: Train AI to ignore gaps or weight them neutrally.
Bias-Checking Prompt
Review our screening process for potential bias.
Our current criteria:
[List your criteria]
Check for bias related to:
- Age (graduation years, decades of experience)
- Gender (activities, language patterns)
- Race/ethnicity (names, organizations)
- Socioeconomic status (schools, companies)
- Disability (employment patterns)
- Family status (employment gaps)
For each potential bias identified:
- Explain how it could affect screening
- Suggest a more neutral alternative
- Propose safeguards to implement
Blind Review Setup
Help me set up a blind resume review process.
Information to remove or obscure:
- Candidate name
- Photos
- Address
- Personal email (if identifying)
- Graduation years
- Dates of employment (keep duration)
Information to keep:
- Skills
- Experience descriptions
- Achievements
- Relevant certifications
Create a template for how anonymized resumes
should be formatted.
5.5 Comparing Candidates
Creating Comparison Frameworks
Create a candidate comparison framework for [Job Title].
Candidates to compare: [Number]
Comparison dimensions:
1. Technical qualification match (required skills)
2. Experience relevance (industry, role type)
3. Achievement indicators (impact, leadership)
4. Growth potential (trajectory, learning)
5. Culture indicators (values, interests)
Format as a comparison matrix I can fill out
for each candidate, with space for notes and
an overall recommendation section.
Side-by-Side Analysis
Help me compare these two candidates for [Job Title].
For each candidate, I'll provide their resume summary.
Candidate A:
[Paste summary or key points]
Candidate B:
[Paste summary or key points]
Compare on:
- Strongest qualifications for the role
- Potential concerns or gaps
- What each brings that the other doesn't
- Questions to ask each in interview
Do not make a recommendation—provide analysis
for my decision.
Trade-off Analysis
Help me think through this hiring trade-off.
Role: [Job Title]
Candidate A strengths:
[List strengths]
Candidate A concerns:
[List concerns]
Candidate B strengths:
[List strengths]
Candidate B concerns:
[List concerns]
Questions to consider:
- Which concerns can be addressed with training?
- Which strengths are most important for this role?
- What are the risks of each choice?
- What would you want to clarify before deciding?
5.6 Efficient Screening Workflows
The Tiered Approach
Tier 1: Knockout Screening Automatic disqualification for missing essentials:
- Work authorization
- Required certifications
- Minimum qualifications
Create a knockout screening checklist for [Job Title].
Absolute requirements:
1. [Requirement 1]
2. [Requirement 2]
3. [Requirement 3]
If any are missing, the candidate is automatically
moved to rejection queue.
Format as a simple yes/no checklist.
Tier 2: AI-Assisted Review AI extracts and summarizes for human review:
- Match against requirements
- Highlight key qualifications
- Flag areas to explore
Tier 3: Human Evaluation Recruiter/manager reviews AI output:
- Makes pass/fail decisions
- Identifies top candidates
- Plans interview questions
Workflow Automation
Design a screening workflow for high-volume recruiting.
Volume: [X] applications expected
Timeline: [X] days to screen
Resources: [X] recruiters available
Create a process that:
1. Handles initial knockout screening
2. Prioritizes most qualified candidates
3. Ensures consistent evaluation
4. Includes bias checks
5. Moves candidates to interview efficiently
Include time estimates for each step.
5.7 Documentation and Compliance
Why Documentation Matters
- Legal protection (especially for federal contractors)
- Consistency across recruiters
- Appeals and disputes
- Process improvement
Documentation Template
Create a screening documentation template for
compliance purposes.
Include fields for:
- Candidate identifier
- Position applied for
- Date of review
- Reviewer name
- Criteria evaluated
- Scores/ratings given
- Notes/rationale
- Decision and reason
- Any AI tools used and how
Also include:
- Instructions for completion
- How long to retain records
- What NOT to document (protected info)
AI Usage Documentation
Create a policy for documenting AI use in screening.
Include:
- When AI assistance was used
- What inputs were provided to AI
- What outputs AI generated
- How human reviewer used/modified AI output
- Final decision and human rationale
This should demonstrate that AI assisted but
did not replace human judgment.
Module 5 Summary
Key Takeaways:
-
Define criteria first: Establish clear, job-related criteria before reviewing any resumes.
-
Use AI as an assistant: AI extracts and summarizes; humans make decisions.
-
Actively combat bias: Check criteria, blind reviews, and audit outcomes.
-
Compare consistently: Use structured frameworks to compare candidates fairly.
-
Document everything: Maintain records for compliance and improvement.
-
Never automate decisions: AI informs, but humans decide.
Preparing for Module 6
In the next module, we'll tackle one of the most challenging HR tasks: writing rejection emails. You'll learn to:
- Write rejections that maintain employer brand
- Be kind but clear with candidates
- Handle different rejection scenarios
- Create templates for efficiency
Before Module 6:
- Review your current rejection email templates
- Think about rejection emails you've received
- Note what makes a rejection feel respectful vs. dismissive
"Fair screening is about giving every qualified candidate a genuine chance—AI can help us do that at scale, if we use it responsibly."
Ready to continue? Proceed to Module 6: Rejection Emails.