AI in HR: The promise, the problem, and the path forward
Leaders like Josh Bersin are touting AI as the solution in HR. Here’s how they are wrong.
Artificial Intelligence is transforming human resources. Thought leaders like Josh Bersin are championing AI as a panacea, predicting that we may no longer need traditional assessments. The narrative is seductive: AI can scan resumes, mine job history, and automate the matching of candidates to roles with unprecedented speed.
But here’s the truth that no one wants to say out loud: AI is only as good as the data it consumes. And in HR, that data is broken.
AI inference engines today rely heavily on resumes, job descriptions and historical performance data to make predictions. But what if the data is biased, outdated, incomplete—or just plain wrong? Then the inference is wrong. The promise of precision becomes a prophecy of error.
The mirage of "smart" hiring
1. Resumes and Job History Are Not the Truth
Let’s start here: resumes are notoriously unreliable. A 2020 study by Checkster found that 78% of job seekers lie or consider lying on resumes. Even when they're accurate, resumes only reflect what someone has done—not what they can do. They offer a narrow, moment-in-time snapshot of someone's career, excluding hidden skills, aspirations, and capabilities that never made it onto the page.
2. Performance Reviews Are Biased
Performance reviews are riddled with bias. Research from McKinsey and LeanIn.org reveals that women and people of color consistently receive less constructive feedback and more vague, personality-based feedback than their white male counterparts. When AI mines this flawed data, it doesn’t eliminate the bias—it scales it.
3. Self-Evaluations Are Inherently Flawed
A 2018 study in the Harvard Business Review found that most people are poor judges of their own capabilities. Some overestimate. Others underestimate. And all too often, interpretation differs: what one person calls “leadership,” another may see as “collaboration.” Without structured definitions and validation, self-assessments are not data. They’re noise.
The dangerous loop of inference
When AI recommends a job based on what a person has done, it assumes that:
They were good at that job.
They enjoyed that job.
They want to keep doing similar work.
They had all the necessary skills to perform that job successfully.
But what if they were underperforming due to a mismatch in strengths? What if they hated that job? What if they want a fresh start?
AI doesn't ask those questions. It just infers. And inference without understanding leads to flawed recommendations. The loop becomes self-perpetuating, reinforcing bad matches and missed potential.
Garbage in, Garbage out: Why clean data matters
AI thrives on data that is:
Complete
Accurate
Unbiased
Interpretable
Unfortunately, most hiring data doesn’t check any of those boxes.
🚫 Job postings? Outdated and inflated.
🚫 Resumes? Incomplete and error-prone.
🚫 Performance data? Subjective and biased.
🚫 Skills data? Often inferred, not measured.
To build predictive models that work, we need to stop feeding AI bad fuel. We need clean, rich, validated data that reflects not only what a person has done—but what they can do.
The solution: Better data for AI
That’s where solutions like MyInnerGenius® come in. Instead of relying on resume-centric inference, MyInnerGenius collects structured, bias-reduced data on a person’s:
Cognitive abilities
Mindsets
Interests and values
Personality traits
Transferable and durable skills
This isn’t about testing knowledge—it’s about uncovering potential. By understanding the “hidden genius” in each individual, MyInnerGenius creates a rich profile that AI can use to make truly personalized, accurate, and future-facing job recommendations.
This is the clean fuel AI has been waiting for.
Call to Action: Create “clean fuel” before you automate
Before HR leaders replace assessments with inference, they need to ask a tough question: "Do we really know this person—or are we guessing based on a flawed mirror?"
The future of AI in hiring isn’t less data. It’s better data. It’s clean fuel, free from bias.
Let’s stop automating flawed systems and start building systems that recognize each person’s full potential. Let’s invest in tools that illuminate what job seekers can become, not just what they’ve done. Only then will AI move from risky inference to responsible intelligence.
About the Author:
David Leaser is an award-winning strategist, C-Suite consultant & program lead in L&D and HCM, Vice President at MyInnerGenius.. He is the founder of the IBM Digital Badge program, a leading-edge digital credential program, the IBM New Collar Certificate Program and IBM’s first cloud-based embedded learning solution. David was a senior strategist for IBM’s Smarter Workforce and the Global Skills Initiative. David is a Commissioner for The RSA (The royal society for arts, manufactures and commerce)'s Digital Badge Commission, a member of the 1Edtech Board advisory group for digital credentials, the national Credential As You Go Advisory Board and a senior advisor to New Markets Venture Capital Group. He provides guidance to the US Department of Labor and the US Department of Education as an employer subject matter expert.
David was appointed as an Industry Fellow in the Center for the Future of Higher Education & Talent Strategy in the College of Professional Studies at Northeastern University, an American Tier 1 university. He is the author of thought leadership white papers on talent development, including “Migrating Minds,” “The Social Imperative in Workforce Development” and Wiley’s “Connecting Workplace Learning and Academic Credentials via Digital Badges.”
David holds an M.A. in Communications Management from USC’s Annenberg School and a B.A. in Communications from Pepperdine University. Connect with David through LinkedIn.