Research & Insights

Presented by Wright Technical Services
Written by Denise D’Angelo

Hiring has always involved some guesswork about a candidate’s future performance. What has
changed is when and where decisions are made. Automated screening tools narrow candidate
pools before meaningful human interaction occurs. For most organizations, the screening
process has become the hiring decision. Strengthening hiring decisions, especially for AI-enabled
roles, requires moving beyond automated filters to understand how candidates
problem solve, explain their reasoning, and work alongside AI.

Screening Limitations

Automated screening identifies fluency and consistency reliably but cannot detect how people
think through uncertainty. Screening now extends beyond resumes. Interviews are increasingly
impersonal, conducted by AI or through recorded video and evaluated against fixed criteria
designed for consistency at scale.

This high-stakes, one-take format can be especially stressful for candidates who reason
differently. These candidates refine answers as they go and adjust assumptions mid-response.
In real work, especially AI-enabled work, iteration is an asset. In automated screening, it
becomes grounds for rejection.

Given approximately 83% of U.S. companies are now using AI resume screening, these patterns
compound across the hiring landscape. I’m seeing similar patterns in conversations with
manufacturing leaders grappling with AI-assisted design validation and insurance executives
evaluating underwriters who work with predictive models.

Automated screening brings clear value: efficiency at scale, consistent initial evaluation, and
reduction of some human biases. The limitation is not the technology itself but relying on it as
the sole filter for capabilities it cannot measure.

The Cost of Mismatches

Misalignment between a hire and the role typically becomes visible within the first few months,
when the work moves beyond predictable scenarios. Problems surface in judgment calls,
collaboration, and knowing when to question AI-generated outputs. For example, a
manufacturing engineer presents AI-generated optimization models to leadership without
validating assumptions against actual plant conditions, leading to a costly failed
implementation; or a developer implements AI-generated code with incorrect assumptions,
breaking downstream integrations.

The U.S. Department of Labor estimates that a bad hire costs approximately 30% of the
employee’s first-year earnings. When screening tools filter for polished answers but the work
requires discernment and adaptability, qualified candidates can get filtered out while mis-hires
may pass automation and underperform in complex projects.

These evaluation challenges extend beyond external hiring. Companies facing hiring freezes or
limited headcount are identifying internal candidates for AI-enabled roles using similar
screening approaches, assessing technical credentials while missing critical capabilities for AI
partnership.

Evaluating for Judgment

In my role running large data teams, we prioritized scaling quickly while committing to evaluate
candidates across many dimensions. When hiring key roles such as data scientists, that meant
moving beyond resume screens and coding tests. Candidates received a real business problem
involving customer behavior data with known quality issues and were explicitly asked to use AI
tools. During live follow-up sessions, we introduced unpredictable variables, questioned
validation methods, and examined how and where candidates overrode AI suggestions.

Strong candidates walked through their reasoning, identified and addressed data quality issues
proactively, and clearly distinguished between AI-directed work and areas requiring human
judgment. Weaker candidates produced polished outputs but struggled to explain results or
adapt when constraints were introduced, exposing the use of AI output without true
understanding.

This approach revealed how candidates think through ambiguity. Once hired, those who
demonstrated critical thinking in interviews adapted well when demanding projects didn’t go as
planned. This method requires more time upfront, yet turnover from mis-hires costs far more
than thoughtful evaluation.

This deeper evaluation doesn’t replace screening, it complements it. Use automated tools to
handle volume efficiently, then invest additional time in roles where judgment and AI
partnership are critical. Not every hire needs this level of assessment, but key technical and
leadership roles benefit from evaluation that goes beyond what screening can measure.

For your next critical hires, consider giving candidates a messy problem, allowing AI tools, and
exploring their results and reasoning. This approach applies equally to internal mobility. As
companies redeploy talent into AI-enabled roles, evaluation can extend beyond current role
performance or technical credentials.

Questions that reveal thinking patterns and AI partnership:

• Where are you responsible for validating or correcting AI outputs, and how do you
  approach that?
• Which parts of your role have changed with AI?
• How has AI increased your capacity, and what do you do with the =me or bandwidth it
  creates?

Measuring Effectiveness

In AI-enabled roles, automated screening rewards consistency and polished responses. The dayto-
day work requires additional competency in reasoning through ambiguity and adaptability to
work with AI.

AI enablement has changed how we work and continues to improve hiring tools. As screening
technology evolves to better assess reasoning and problem-solving, the principle remains:
validate that your tools surface what the work requires. Track which candidates pass
automated screening, receive interviewer recommendations, and flourish six months later.
When these align, your evaluation works. When they don’t, both screening criteria and
interview approaches warrant refinement to strengthen hiring decisions.

About Denise D’Angelo

Denise D’Angelo is a transformation executive specializing in AI workforce readiness, governance, and operating model evolution. Her experience spans Fortune 500 financial services, research institutions, and defense programs, where she has led enterprise transformation and AI-driven data strategy in regulated environments. A trusted executive facilitator and consensus builder, she brings disciplined clarity to workforce design, decision authority, and role strategy, aligning expanded AI capability with accountability and measurable performance, grounded in a Master’s degree from Johns Hopkins’ Whiting School of Engineering and graduate studies in Data Privacy Law at Seton Hall.