AI Just Shortlisted Your Next Employee. Did Anyone Ask HR?
Why is artificial intelligence in hiring exciting, dangerous, and everything HR needs to be paying attention to right now?
By Rasiah Keerthana | MBA Batch 37 | Jul 2026
The Algorithm Does Not Know Your Student’s Story
Platforms like ApplyBoard already use AI to screen student profiles
and match them to university requirements. It is fast and removes hours of
manual work. But what happens to the student whose overall GPA looks weak, yet
whose individual subject marks tell a completely different story? A human
consultant reads that transcript differently, going beyond the overall GPA to
review individual subject marks, and considering the student’s own Statement of
Purpose to understand their motivation and context. An algorithm flags a weak
number and moves on. A consultant sees a person and finds a way forward. In an
industry where one decision can change someone’s entire future, that difference
is not small.
Where HR Fits in an AI-Driven World
The Ulrich Model positions HR as a strategic partner, employee champion,
and change agent, not an administrator (Ulrich, 1997). That role
has never mattered more. Ninety percent of US employers now use AI screening
tools, most relying on the same few vendors (Stanford HAI, 2026). When
everyone uses the same algorithm, the same candidates get rejected everywhere,
not because they are unqualified, but because the data was never neutral to
begin with. Peer-reviewed research confirms that AI-driven HR systems
consistently replicate biases in their training data, particularly around
gender, race, and socioeconomic background (Sony et al., 2025).
Amazon scrapped its own AI hiring tool after it systematically downgraded
applications from women, not by design, but because it learned from
historically biased data.
Read more: AI Hiring Tools Can Yield Racial Bias and Systemic Rejection - Stanford University
Exciting, Yes… But at What Cost?
AI saves time and removes human inconsistency; that is genuinely
valuable. But the deeper concern is this: AI works with algorithms. Humans work
with empathy, instinct, and the ability to think beyond data. When organizations
depend entirely on AI to evaluate and decide, what happens to their own ability
to judge? If HR professionals cannot evaluate independently anymore, they are
not strategic partners; they are just approving what a machine already decided.
The California Civil Rights Department recognized exactly this risk when it
approved landmark regulations on 27 June 2025, effective 1 October 2025,
requiring meaningful human oversight over any automated employment decision,
making clear that innovation must serve fairness, not replace human judgement (California
Civil Rights Department, 2025). That is not anti-technology. That is
exactly what the Ulrich Model’s employee champion role was designed to protect.
Read more: Research: AI Is Changing What Employers Want from New Hires - Harvard Business Review
AI should be a tool HR uses, not a replacement for the judgement HR
develops. The next generation of consultants needs sharper critical thinking
and stronger human skills, not because AI is the enemy, but because those are
exactly the things it cannot replicate. So here is the question: if AI made the
last hiring decision in your organization, did anyone actually check its work?
References
California Civil Rights Department (2025) Civil Rights Council secures approval for regulations to protect against employment discrimination related to artificial intelligence. State of California. Available at: https://calcivilrights.ca.gov/2025/06/30/civil-rights-council-secures-approval-for-regulations-to-protect-against-employment-discrimination-related-to-artificial-intelligence/ (Accessed: 05 July 2026).
Doucette, J. and Gaur, V. (2026) 'Research: AI is changing what employers want from new hires', Harvard Business Review. Available at: https://hbr.org/2026/07/research-ai-is-changing-what-employers-want-from-new-hires (Accessed: 05 July 2026).
Sony, M.M.A.A., Amin, M.B., Ashraf, A., Islam, K.M.A., Debnath, N.C. and Debnath, G.C. (2025) 'Bias in AI-driven HRM systems: investigating discrimination risks embedded in AI recruitment tools and HR analytics', Social Sciences & Humanities Open, 12, 102082. Available at: https://doi.org/10.1016/j.ssaho.2025.102082 (Accessed: 06 July 2026).
Stanford HAI (2026) AI hiring tools can yield racial bias and systemic rejection. Stanford Human-Centered Artificial Intelligence. Available at: https://hai.stanford.edu/news/ai-hiring-tools-can-yield-racial-bias-and-systemic-rejection (Accessed: 07 July 2026).
Teamflect (2026) 5 Ways AI Is Transforming HR in 2026 (Recruitment, Analytics, Ethics & More). [Video]. YouTube. Available at: https://www.youtube.com/watch?v=vHiIxvoreco (Accessed: 07 July 2026).
Ulrich, D. (1997) Human resource champions: the next agenda for adding value and delivering results. Boston, MA: Harvard Business School Press.
This article deeply resonates with my belief on empathy over algorithms. I strongly agree with your statement that AI should be used only as a too and not as a replacement for human judgement. To answer your question, the last hiring in my company through AI deprived a deserving candidate only because he was falling short in some numbers. But when interviewed in person, he was more than apt for the job with his visions and experiences.
ReplyDeleteThank you for sharing that example; it's a good illustration of the gap the post is pointing to. The deeper issue isn't really AI versus human judgement; it's what the algorithm was told to score on. If the criteria only capture numbers, it will miss fit every time, regardless of how the process is run. That points to fixing what the tool measures, alongside keeping a human review step for candidates it screens out.
DeleteAlthough the blog highlights valid concerns about AI bias, I believe the issue is not AI itself but how organizations implement and monitor these systems. Human decision-making is also affected by personal bias, so completely relying on humans does not always guarantee fairness. A combination of AI insights and human judgement may create a more effective and balanced recruitment process.
ReplyDeleteThank you, that's a fair point. Both are biased in their own way. The difference is scale: a biased recruiter affects decisions one at a time, while a biased algorithm can repeat the same pattern across every candidate at once, which makes monitoring more urgent, not less necessary. So combining AI and human judgement only helps if the combination includes real auditing, not just having both involved.
DeleteMost of the organizations always need a quick solution for everything. So when hiring time & after that they are not checking properly that decision is correct or not until person get failed on their tasks. So my opinion is when we are hiring someone we need to use Al as a tool & there should be a human engagement as well.
ReplyDeleteThank you, that's a good observation, and it points to a timing problem more than a tool problem. Waiting until someone fails to check whether a hiring decision was right means the cost has already been paid for both the employee and the organization. A short structured check-in during the first few months would likely catch a bad match far earlier than waiting for failure to reveal it.
DeleteA brilliant and timely piece, Keerthana! Positioning the Ulrich Model’s Employee Champion role as a safeguard against algorithmic bias is a spot-on application. However, offering a subtle counter-perspective: romanticizing 'pure human judgment' can be equally dangerous. Human recruiters carry unconscious affinity bias, fatigue errors, and subjective gut-feelings that are far harder to measure or audit than an algorithm. The strategic solution isn't resisting AI shortlisting in favor of manual reviewing—it is Algorithmic Calibration. Instead of HR acting as a manual filter after the AI, HR must act as the systemic auditor before the algorithm runs, setting unbiased parameters and auditing rejection patterns. True strategic HR doesn't replace AI; it governs it.
ReplyDeleteThank you; "governs it" is a good way to put it, and I agree pure human judgement isn't the safer default it's often assumed to be. I did only add that calibrating parameters before the algorithm runs handles bias at the start, but real-world data shifts over time, so rejection patterns still need auditing after launch, not just once at setup. Governance probably has to be ongoing, not a single upfront step.
DeleteExcellent post. The transcript example makes the point better than any statistic a weak overall GPA hiding strong subject marks is exactly the kind of story an algorithm can't read.
ReplyDeleteYour observation that most employers rely on the same few vendors is the part I hadn't considered. If everyone screens with the same tool, a rejection stops being one company's decision and becomes a closed door everywhere. Sharp writing.
Thank you; that vendor point is worth pushing a bit further, too. When the same tool rejects a candidate across many companies, the bias doesn't just narrow their chances; it also becomes very hard to trace back to any single decision-maker. That's arguably the bigger risk: not one wrong call, but a pattern nobody can be held accountable for.
DeleteThe discussion on context and empathy is valuable. While AI can analyse data and identify patterns, understanding employee emotions and organisational culture requires human insight and experience.
ReplyDeleteThank you, I agree, though I did add that human insight only helps if it's actually brought into the decision itself, not just treated as a general capability HR has somewhere. If empathy and context aren't built into the actual hiring step, having the capacity for them doesn't change the outcome.
DeleteGreat perspective on AI recruitment. Combining technology with human judgment ensures fair, efficient, and effective hiring decisions for organizations.
ReplyDeleteThank you; I did frame "ensures" a little more carefully, though. Combining the two only leads to fairness if there's clarity on which one wins when they disagree, since efficiency and fairness don't always point in the same direction. Without that, "combining" can just mean AI decides and a human signs off.
DeleteThis comment has been removed by the author.
ReplyDelete