Some of you may have noticed the terms candidate relevancy or candidate match creeping into ATS application processes. Specifically, the ADP ATS has introduced the Artificial Intelligence (AI) based calculation of a Candidate Relevancy Score and applicants need to opt in or out of the process. The Workday ATS and others have similar improvements planned. Should you allow AI-powered scoring of your resume or opt-out? Let's break it down.
What is Candidate Matching?
Candidate matching is the degree to which a candidate profile aligns with the specific requirements of an open job and the process by which recruiters identify suitable candidates from a pool of applications. The process has been around since the 90s.
Talent professionals additionally use other factors like soft skills, cover letters, experience, education, soft skills, skills tests, coding challenges, and video interviews to assist in their assessments. Each company uses a slightly different algorithm to assess candidate match. For example, Microsoft might require a coding challenge for computer programmer positions, and Amazon may not.
The Job Search Landscape is Changing with AI Screening
About 15 years ago, Applicant Tracking Systems introduced capabilities for manually searching candidate databases by identifying and searching relevant keywords and skills. The advent of AI and machine learning has slowly started to change the landscape of how candidate profiles are sourced and assessed. A few ways talent selection is changing:
Assessing candidate relevancy scores and match prediction for interview selection
Use of AI screening tools and AI systems
Social media screening for red flags
Evaluating video interviews for job match assessment
Use of personality or aptitude tests to assess cultural fit.
How can Large Language Models (LLMs) Impact the Resume Review Process?
LLMs process the nuances of language, assessing the meaning and relationships between words and concepts. They can also interpret complex information not explicitly stated in a resume. Therefore, using LLMs should provide a more finely tuned Candidate Relevancy Score than just keyword and skills matching.
LLMs can significantly enhance the accuracy and efficiency of resume screening by understanding context and intent. For example, they can distinguish between similar job titles with different responsibilities or recognize industry-specific jargon and acronyms. This deeper understanding enables the AI to more accurately match candidates to job requirements, reducing the likelihood of overlooking qualified applicants due to semantic nuances. LLMs are steadily getting better at evaluating resumes.
LLMs are learning from vast datasets, continuously improving their ability to predict candidate success based on previous hiring outcomes. They can also identify soft skills and personality traits inferred from the language used in resumes, providing a more holistic view of a candidate's potential fit within a company's culture. This advanced analysis goes beyond surface-level matching. ATSs are just starting to roll out new capabilities.
How can ML Enhance the Candidate Screening Process?
Machine learning uses historical data to predict the likelihood of candidate success in a role based on previously identified patterns and relationships in past hirings. These predictive learnings adjust the Candidate Relevance Score.
For example, a job searcher applies for a Customer Service position. It requires excellent communication, problem-solving ability, and experience working in a fast-paced environment.
The LLM scans the resume for keywords related to communication skills, such as conflict resolution.
The ML model analyzes past hiring data and identifies patterns, like a higher success rate for candidates with previous call center experience.
Resumes showing customer service ability not captured by ML-identified keywords or without traditional call center backgrounds would receive lower Candidate relevance scores.
The theory is that the ML model will learn and improve over time to use more finely tuned keywords, and recruiter reviews will catch and include resumes with lower scores.
How Does the Applicant Tracking System Assign a Candidate Score?
Today, most ATSs do not leverage machine learning to assign a candidate score, but many companies are assessing their future use. Following are the processes the ATSs follow with and without applied machine learning.
Assigning a Candidate Score Without Machine Learning
Submit Resume - The candidate submits their resume to the ATS database, which serves as the initial point of contact between the applicant and the hiring organization.
Assign Non-AI Relevancy Score - The ATS uses explicit keywords like job titles, skills, and educational background to assign a candidate relevancy score and rank applicant resumes against the job they apply to by using logical search terms. No AI-type logic is used.
Recruiter Review - Recruiters use manual resume reviews or non-AI candidate scoring to prioritize candidates for further review and interview selection.
NOTE: Many recruiters do not currently utilize candidate scoring due to inconsistent results.
Assigning a Candidate Score Leveraging Machine Learning
Obtain Candidate's Approval - After resume submission, the candidate is asked to opt in or out of the candidate scoring process.
Extract Concrete and Contextual Information - The LLM goes beyond basic keyword matching by understanding the meaning and relationships between words and concepts. This allows it to interpret complex information that may not be explicitly stated in the resume, such as industry-specific jargon, project outcomes, and contextual job roles.
Extract Information - The Large Language Model (LLM) processes the resume by extracting all relevant and contextual information, such as inferred skills, experience levels, and language cues.
Apply Predictive Modeling - The Machine Learning (ML) model then analyzes the extracted data, leveraging specifically identified data to predict the candidate's potential success in the applied role. This includes assessing how previous candidates with similar profiles performed in similar roles, considering both hard and soft skills
Develop a Candidate Relevancy Score - The insights from the LLM and ML models are combined to produce a final candidate relevancy score. This score reflects how well the candidate's profile matches the job requirements, reflecting both the quantitative data (such as years of experience and qualifications) and qualitative data (such as inferred competencies and cultural fit).
Recruiter Review and Prioritization - Recruiters use these scores to prioritize candidates for further review and interviews. Due to potential bias, recruiters carefully review candidate records and scores as part of the candidate identification process.
Note: Companies are just starting to use AI to assign candidate scores. Each company may build slightly different models and should be reviewed for bias testing.
How ADP Assigns a Candidate Relevancy Score
ADP first asks candidates who have applied to other jobs to opt-in to the review process.
ADP uses large language models (LLM) and Machine Learning (ML) to assess how well a candidate's education, skills, and experience match a posted job.
Each of the three areas receives a separate score, which is then weighted and summed. Weighted adjustments are assigned by the company and not shared with applicants.
ADP states that the Candidate Relevancy score will not replace recruiter reviews or final candidate decision-making.
ADP additionally shares statistical evidence of no gender or ethnicity bias.
What are AEDTs?
Automated Employment Decision Tools
A system that uses artificial intelligence, machine learning, algorithms, machine learning, or other computational methods to assist the decision-making process for any employment actions (hiring, promotions, layoffs, job assignments. Examples of AEDTs are:
ATSs that evaluate and prioritize applications through keyword analysis of resumes.
Chatbots that ask job candidates about their qualifications.
Are there Laws that Protect Job Seekers from AI Bias?
AI Bias Protection Laws
NYC passed Local Law 144 of 2021 regarding automated employment decision tools (“AEDT”). The law prohibits employers/employment agencies from using an AEDT unless the tool has passed a bias audit within one year of using the tool. Information about the bias audit must be publicly available, and notices provided to job candidates and employees.
The EEOC has published federal guidance to companies on how to audit their ATSs (even if 3rd party tools like Workday) for bias and discrimination.
States are beginning to pass laws that regulate the use of AI in hiring and other employment decisions. Check your state!
Companies are choosing to add disclaimers and informed consent in their hiring processes. An example is ADP's Candidate Relevancy FAQ.
Companies such as Unilever, Hilton, Siemens, Pepsi, IBM, Intel, PWC, and Accenture are leveraging AI in their employment processes. A few areas include: eliminating hiring bias and increasing diversity, leveraging chatbots for pre-screening, candidate matching/ranking, video interviewing, removing biased language in job descriptions, matching candidates to internal job openings, machine learning-based candidate assessments, and more.
What is Talent Rediscovery?
ATSs have long offered recruiters the ability to manually search their resume database for candidates who have applied to other jobs that may meet newly posted positions. This process is called talent rediscovery. Employers are now using ATSs to leverage AI to automate the previously manual talent rediscovery process. Candidates are not asked whether they consent to this resume evaluation process.
How Do I Participate in Talent Rediscovery?
Some recruiters search their ATS for existing candidate records against newly open positions. Candidates should target at least 10 companies and apply to existing positions so their resumes are in the ATS and considered for future open positions.
Job seekers need to identify employers that allow candidates to upload their resumes to their ATS without applying for a job.
Do All Recruiters Leverage Talent Rediscovery?
Not all recruiters search for existing talent - Many recruiters still rely on manual searches or do not focus on candidate rediscovery searches.
Not all ATSs offer AI-automated screening - Some companies use older ATSs that do not have this capability.
Industry Practices - Some industries rely more heavily on AI screening, while others may prioritize human judgment. Researching the hiring practices of your target companies can provide insights.
Should I Opt Out of AI Resume Screening?
Opting In
Streamlined Screening Process - Opting in can save time and effort as the recruiter may not require a new resume (as one is already in the ATS).
Feedback and Position Matching - AI-driven screening provides instant feedback on how well a resume matches the newly posted job description to recruiters.
Exposure to Selection Bias - While opting into an AI-based candidate relevancy score, it is influenced by algorithms and data inputs that may include unseen biases, which can impact the evaluation process. See the section on AEDTs and current bias protection laws.
Transparency - Not all companies disclose the exact process and factors that affect relevancy scores in AI systems. This lack of transparency can make it challenging for candidates to understand why their application is rejected.
Data-Driven Decisions - The assessment of candidate relevance in AI screening is heavily data-driven, relying on patterns and historical data more strongly than human intuition. New laws require human review of any AEDT system that uses AI for candidate scoring, but this area is still evolving.
Opting Out:
Bypassing AI Bias - By opting out, candidates avoid potential biases introduced by AI-based algorithms. This can be advantageous for those with unique skills or experiences that may not align perfectly with typical keyword matches.
Human Review Advantage - Resumes that highlight unique skills and experiences may be better received by human recruiters. Human reviewers are often more adept at recognizing value beyond data points, such as creativity, leadership potential, and cultural fit.
Screening Process - The screening steps for opting out of candidate relevancy scoring would be more manual and could take longer, putting the application in jeopardy.
In conclusion, whether to opt-out of AI resume screening depends on an ATSs capabilities, industry standards, recruiter practices, and whether you are comfortable with allowing an AI algorithm to evaluate your resume prior to any recruiter review. Weighing these factors can help you make an informed decision that maximizes your chances of landing the job.
AI-Powered Candidate Scoring Suggestions
Job searchers strongly concerned about bias or lack of company transparency around the hiring process may want to opt-out.
Suggest opt-out if the candidate review process only utilizes AI (and no human review).
Suggest opt-in if the resume is a clean match and the job searcher wants increased visibility.
Suggest opt-in if the state/municipality/city where you would be working has anti-bias laws.
As AI models are trained and machine learning improves its models, the results should improve and bias risks reduced, but only time will tell.
Is your job search feeling unfocused? Is your resume lacking clarity?
I’ll work one-on-one with you to translate your technical skills and achievements into business value. Together, we’ll build a compelling story and LinkedIn profile that highlights your impact, catches the attention of recruiters and hiring managers, and provides the insights you need for better results! Here’s how I can help:
➡️ Book a free 30-minute consultation to discuss your career needs and goals
➡️ Grab your free job search guide to learn to network like a pro and accelerate your job search
➡️ Bonus: Get your Is LinkedIn Premium Worth It? analysis to make the best decision for your career