.png){kind=logo}
By Madhusudhana Rao Podila, CPTO, Troogue.ai
An article published recently in the Harvard Business Review raised concerns about how Generative AI was “rapidly undermining the reliability of traditional hiring signals, making it easier for candidates to manufacture polished résumés and perform convincingly in remote interviews with real-time assistance tools.” The resulting concern for hiring managers is how to distinguish genuine capability when every candidate, at least on paper, appears to have the same skills and competencies.
Generative AI has become a tool in the hands of candidates to create polished resumes and present themselves as an ideal fit for almost any role. The result is a sea of sameness, making it increasingly difficult for recruiters to determine the right fit for a role. This is not to say that Generative AI is the culprit here - far from it. However, rather than using it to mask genuine capability, candidates should use it to communicate and extend that capability.
A resume right now has become more of a statement of claims rather than proof of a candidate’s capabilities. Capability also includes what a person can and cannot currently do, how they reason through a problem, and how they validate the outcome. It’s not just about the end results now but also about processes followed to reach the desired outcome. Quite similar in fact to solving a mathematical equation where marks are awarded not only for the correct answer but also for the steps followed to solve that equation. In certain companies for instance, observed performance is used to determine candidate suitability by evaluating them on their approach to resolving a problem, the different options they considered including use of AI, how they handled feedback and constructive criticism and finally, their conviction for the solution they recommended.
Given the all-pervasive nature of AI in workplaces, assessments need to measure not just a candidate’s AI-assisted performance but also their independent or baseline capability. Too much reliance on AI can have an adverse impact on performance, as a recent study by Anthropic found out.
In a controlled trial of software developers learning a new skill - in this case, a Python library - it was found that developers who used AI tools scored 17% lower on a quiz covering conceptual understanding, code reading and debugging than those who coded manually, even though developers using AI completed their assignment slightly faster. The speed difference was not statistically significant.
The study, which had a dual objective, also revealed another interesting result - that more than the use of AI, it was how the developers used AI which determined their final scores. Participants who scored higher didn’t just use AI to generate code but rather, they “used AI assistance… to build comprehension while doing so—whether by asking follow-up questions, requesting explanations, or posing conceptual questions while coding independently.”
For practical implementation, recruiters and hiring managers can assign real-life tasks or simulations to candidates and ask them to solve comparable exercises with and without AI tools. This dual-mode assessment can more accurately determine a candidate’s knowledge and skill sets as well as how they use AI. It can also show whether the candidate can challenge generated output, detect errors and remain accountable for the final recommendation.
Hiring managers look for verified capability rather than just a badge or a certification listed on a resume for any given skill set. Verified capability is based on results of realistic assessments - such as a coding test with real bugs to fix or a team challenge that requires collaboration across functions to see how an individual works in a multicultural environment. Such evidence makes capability substantially harder to overstate.
For example, a candidate may list experience in designing scalable systems. A realistic exercise can reveal whether that person can clarify ambiguous requirements, make appropriate architectural trade-offs, identify failure scenarios and defend the proposed design when assumptions change.
Verified capability thus requires authentic context in which tasks reflect real-life constraints and client requirements; process visibility, wherein a candidate’s step-by-step approach is evaluated; output evaluation to determine the quality of work delivered and whether it is complete; and structured feedback on each of the preceding aspects. It should also reveal what the person can do independently, where AI adds value and when further expertise is required.
Such evidence must be handled responsibly, with explainability, privacy, consent, bias monitoring and human oversight.
The Digital Skills Passport, introduced by Lisbon City Council in 2017, offers an early example of maintaining an evolving record of digital competence. It is not equivalent to a verified-capability model, but it illustrates the shift from static credentials towards records that develop as people learn. A capability passport can take this idea further by showing evidence of a candidate’s proven skills and being updated as new capabilities are demonstrated.
Since these digital documents will have not only a candidate’s structured assessment data and skill scores but also a holistic assessment of their capabilities, recruiters can choose profiles based on their requirements - narrowing the haystack to find the needle, so to speak. Resumes, in a sense, can act as introductions for a candidate, whose fit for a particular role will be demonstrated through verified competencies rather than what is claimed. We are building towards a Verified Capability Passport that brings together assessment results, work evidence, independent and AI-assisted performance, and areas for development.
The résumé may remain the introduction. Verified capability will increasingly determine trust.