Feature
Workers are increasingly disillusioned by AI’s role in hiring and promoting, lamenting its role in ruling out qualified candidates. Many feel they cannot advance or find work without knowing someone to hopscotch them past these screening machines.
The issue goes beyond candidates. Companies also risk missing out on high-potential candidates who get categorized by a narrowly focused or poorly programmed algorithm.
“AI is portrayed as either a panacea or a potential risk,” said James Dyson Jr., founder and CEO of Think Evolutionary, a fractional talent acquisition firm.
This tale of two realities is increasing confusion around AI’s decision-making capabilities — and its role in talent management.
AI Is Overestimated and Misunderstood
Many of us tend to overestimate AI, expecting it to be able to assess soft skills like empathy or leadership with the same accuracy as humans. But this sort of nuance escapes AI’s detection, says Shail Khiyara, founder and CEO of Plutoshift AI.
“Most AI systems are designed to mitigate bias by focusing on objective criteria like skills and experience,” Khiyara said.
The idea that AI can do more, or at least do it to the level of humans, may stem from what Michelle Duval, founder and CEO of Marlee, calls a “lack of AI literacy.” Non-experts can’t keep up with AI’s rapid evolution, which creates “a gap between perception and reality,” Duval said. In turn, this perception problem results in people viewing AI as an adversary to “get around.”
The growing number of very public examples calling out AI’s impartiality doesn’t help. As Dani Herrera, talent and DEI consultant at Key & Partners Talent Management, reminds us: “Just last year, we saw how popular AI platforms gave preference, unprompted, to male candidates over female candidates and disqualified candidates based on what school they went to.”
And there have been several other instances of AI completely disregarding equity and accessibility.
All this ultimately raises the level of fear over what would happen if (or when) AI replaces humans in certain tasks — particularly so in talent management. “AI should result in reskilling and upskilling opportunities that enable workers to transition to more strategic and rewarding roles rather than simply eliminating positions,” said Khiyara.
Humans will need to determine how AI augments man-made models, and ethically expediting decision-making is a critical use case.
Related Article: Use Generative AI to Write Performance Reviews? Not So Fast
Ethically Expedite Decision-Making
Dyson believes there is a lot of work left to be done before we can fully rely on AI. “Humans must implement AI with strong ethical and compliance standards to avoid amplifying biases,” he said, adding the data used to train AI must be free from historical biases, or it will reinforce those biases in its decision-making.
This applies to small and mid-sized businesses as much as enterprises, as AI-driven tools are becoming increasingly available “to organizations of all sizes, democratizing access to sophisticated talent management capabilities,” Duval said.
An objective AI-driven system therefore should:
- Ignore demographic information that might lead to conscious or unconscious bias, such as age, gender, race or educational background, and focus solely on relevant qualifications, skills and experience.
- Analyze a broader range of data points and uncover talented candidates or promotion-worthy employees who might be otherwise overlooked.
- Standardize the evaluation process, applying the same criteria consistently across all candidates.
Herrera added that when properly and objectively programmed, AI could also help people from historically excluded communities receive more equitable feedback.
And it can self-improve.
“AI can track decision-making patterns over time, highlighting any biases that may still slip through, enabling companies to course correct,” shared Khiyara. However, AI analyses are not always clear-cut, which brings us to the sticky prospect of adapting to opaque AI.
Related Article: Talent Experience Is Broken and the Technology Isn't Helping
Acknowledging Opaque AI Strengths & Challenges
Opaque AI systems — those where the decision-making process isn't entirely transparent — handle large volumes of data quickly and efficiently, identifying patterns that may be too complex or time-consuming for human recruiters to notice. They cannot be swayed by a candidate’s charisma or appearance.
With opaque AI, companies might avoid overly subjective decisions that could lead to bad hires or missed promotion opportunities.
Duval said AI’s opacity could also prevent candidates from gaming the system, and opaque systems might allow companies to focus more on the quality of the AI's decisions rather than debating the intricacies of its process.
Conference
Sep
16
HR Tech Conference Las Vegas 2025
Register
Conference
Oct
27
Gartner HR Symposium/Xpo Orlando 2025
Register
“However,” Dyson cautioned, “it can also be a double-edged sword.” When decisions aren't transparent, it becomes challenging to ensure they are fair and free from bias. It can also undermine trust among candidates and employees if they don't understand why they were not selected for a role or promotion.
Still, Duval said, the benefits of transparency generally outweigh those of opacity. “At Marlee, we strive to provide clear explanations of how we assess candidates' motivational work styles and values. This approach fosters trust and allows for continuous improvement of our system.”
Related Article: Who's Responsible for Responsible AI?
Creating a Compromise and Forging Forward
All agreed that transparency cannot be an afterthought when using AI to make talent-related decisions. “We know that women, women of color and people from historically excluded communities tend to receive more inequitable feedback purely based on their attitude instead of skills,” Herrera said. “These tools help minimize those biased decisions.”
Transparency supports change and builds trust. But that requires balance.
“All aspects of an AI system do not need to be completely transparent, but companies should ensure they can explain how the AI reached its conclusions,” said Dyson. They can establish governance frameworks, combining efficiency with human oversight to maintain ethical standards and operational effectiveness.
Khiyara says on the AI-powered talent management path, there are other transformations we can expect, including anticipative dynamic learning systems that evolve as quickly as the fields they teach, focused on optimizing personalized learning outcomes and enhancing the user experience for trainers and learners.
These insights are likely to unlock individually targeted professional development trajectories, unearth hidden leaders and execute meritorious and meticulously just promotion strategies — and see employees use this insight to become architects of their own destinies.
And isn’t this the penultimate goal, just short of flying cars, that we all expected when this technological journey began?
