Publication Note

I am developing Organizational Systems Intelligence (OSI) and Professional Identity Architecture (PIA) to help address some of the major challenges in the modern hyper-speed landscape of hiring and job searching. OSI/PIA seeks evidence-bounded, methodologically cautious ways to describe hidden capability at both the individual and organizational level. The goal is to improve visibility into capability and potential without scoring, labeling, or converting people into simplified metrics whose humanity disappears inside modern organizational analysis systems. This is the first article in a continuing public series describing that work as it evolves.

Article

Many organizations describe their hiring problem as a talent shortage. That may be partly true in some fields, regions, and labor markets. But it is not the whole problem. A growing body of research and reporting points to a deeper failure: modern hiring systems often struggle to recognize capability when it does not arrive in the expected format. The problem is not simply that organizations cannot find capable people. The problem is that capability has to pass through a long chain of translation before a person is understood. That chain is not purely technical. It includes software, human bandwidth, timing, recruiter attention, job-description design, communication practices, and organizational assumptions. In other words, the problem is not only whether a machine reads a resume correctly. It is whether the whole hiring system preserves enough signal for interpretation.

  • a resume format;
  • job-posting language;
  • keyword filters;
  • applicant tracking systems;
  • automated screeners;
  • recruiter workload;
  • inherited job descriptions;
  • degree requirements;
  • historical hiring patterns;
  • interview conventions;
  • organizational assumptions about what a “qualified” person looks like.

At every stage, capability signal can be lost.

A person may have the ability to do the work, but not the exact vocabulary the system expects. They may have learned through military service, caregiving, internal process improvement, informal leadership, self-directed technical learning, cross-functional work, or non-linear career movement. They may have built capability in real environments without having a resume that matches the standard template. When that happens, modern, layered hiring systems may not see a person with transferable capability. Hiring systems may see an organization’s best candidates as a mismatch for what the machine-readable part of the system expects. A human in the hiring process may not even know their next top performer exists at all.

The System Is Built to Recognize Familiarity

Harvard Business School’s “Hidden Workers” research describes how hiring processes designed to find “perfect” candidates efficiently can systematically exclude qualified workers, including caregivers, veterans, formerly incarcerated people, people with disabilities, and others whose work histories do not fit conventional patterns.1 That framing matters because it shifts the question from:

“Are there enough qualified people?” to “What kinds of qualified people has the system been designed not to see?” In the HBS and Accenture hidden-workers report, automated talent systems are described as often missing “the perfectly capable.” The report identifies hidden workers as a large group of people whose potential is restricted by the way hiring systems define, filter, and interpret qualifications.2 That is not merely an ethical issue, although it is certainly that. It is also an operational issue. If a hiring process excludes qualified people before human review, the organization does not just harm applicants. It damages its own access to capability. The organization may then conclude that talent is scarce, when part of the problem is that its own systems are filtering talent out.

The Self-Selecting Candidate Pool

A major weakness in many hiring systems is that they do not measure human capability directly. They often measure similarity to candidates and employees who have already passed through the organization’s existing filters. That creates a self-reinforcing loop. An organization defines an ideal candidate based on its current workforce, prior hires, industry norms, job-title patterns, degree expectations, and historical success signals. Automated tools then use those patterns to filter and rank new applicants. The people who survive that process become the next data pool. Over time, the system becomes better at recognizing people who resemble the existing frame, while remaining weak at recognizing capability that developed outside it. This is not only a bias problem in the narrow sense. It is a visibility problem. People with non-linear careers, military experience, caregiving gaps, self-directed technical learning, cross-domain experience, informal leadership, operational repair work, or unconventional educational paths may possess real capability. But if their evidence does not resemble the organization’s preferred pattern, the system may classify them as less relevant before their capability is ever interpreted. The result is a closed reference system: the organization uses its own limited past recognition patterns to decide what future capability should look like. That can make hiring appear rational while quietly excluding people whose skills, judgment, and potential exist outside the familiar organizational or industry frame.

AI Did Not Create the Problem. It Accelerated It.

AI did not invent brittle hiring systems. Keyword filtering, rigid degree requirements, narrow experience matching, and resume conventions existed long before generative AI. But AI can intensify the problem. When automated tools are used mainly to reduce volume, rank applicants, or accelerate rejection, they can make old assumptions operate faster and at greater scale. A flawed human process may reject qualified people inconsistently. A flawed automated process can reject them consistently. That distinction matters. A bad filter at small scale is a nuisance. A bad filter at large scale becomes infrastructure. Secondary AI-governance commentary has identified recurring failure modes in AI hiring systems, including keyword dependency, resume parsing errors, historical-data bias, and over-rigid filters.3 More broadly, reporting on AI adoption suggests that workplace AI confidence can fall when tools are adopted without enough training, context, or support.4 The risk is not only that an AI tool makes a mistake. The risk is that the mistake becomes normalized as part of the organization’s talent pipeline. Responsible use of AI in hiring should therefore begin with a simple principle:

AI should help organizations discover evidence of capability, not merely accelerate exclusion. This is where many current systems appear misaligned. They are often optimized for speed, volume management, and administrative efficiency. Those are real needs. But speed is not the same thing as judgment. Reducing applicant volume is not the same thing as identifying capability. A faster “no” is not automatically a better hiring system.

The New Problem: Manufactured Polish

There is another side to the problem. Harvard Business Review has argued that generative AI is weakening traditional hiring signals by making it easier for candidates to manufacture polished resumes and perform convincingly in remote interviews.5 That creates a two-sided failure. On one side, real capability may be rejected because it is not legible to the system. On the other side, manufactured polish may pass through because it is optimized for the system. That is a dangerous combination. A hiring system that rejects unconventional capability while rewarding machine-optimized presentation is not measuring talent. It is measuring compatibility with the hiring interface. In that world, the resume becomes less a record of capability and more a performance object. The candidate learns to satisfy the scanner. The employer learns to distrust the signal. Both sides adapt to the system rather than the work. That is not a healthy labor market. It is a translation failure with feedback loops.

False Negatives May Be the Hidden Cost

Hiring conversations often focus on false positives: the person who gets hired but cannot do the job. False negatives may be just as damaging: the capable person never reaches the interview, and the organization may never know what it lost. They are harder to see because they disappear from the system. They do not show up as bad hires. They do not create onboarding problems. They do not fail performance reviews. This is especially important for people whose careers do not fit a simple linear pattern. Career changers, veterans, caregivers returning to work, people without traditional degrees, people with hybrid technical and human skills, and people who developed capability through lived operational experience may all be vulnerable to being misread. Their evidence exists, but it is not automatically legible. This is the core issue OSI-PIA is being developed to examine.

The Resume Is Not Enough

The traditional resume asks people to compress a complex human development history into a narrow format:

  • job title;
  • employer;
  • dates;
  • bullets;
  • credentials;

  • keywords. That format can be useful. It is quick. It is familiar. It allows comparison, but it is also lossy. Resumes often struggle to show how capability actually develops across roles. They struggle to represent learning, judgment, trust-building, informal leadership, systems thinking, operational repair, and cross-domain transfer. Many of the most important things people do at work are not easy to express as standard accomplishment bullets:

  • stabilizing a failing process;
  • preserving institutional knowledge;
  • training others informally;
  • translating between technical and non-technical teams;
  • making a system usable for the people who actually depend on it;
  • noticing risk before it becomes visible;
  • improving coordination across departments;
  • building trust after breakdown;
  • converting messy experience into repeatable practice.

These are real forms of capability but they are often under-described, under-measured, or miscategorized. A hiring system that cannot interpret those signals may mistake depth for vagueness, breadth for lack of focus, and non-linear growth for inconsistency.

What PIA Is Trying to Build

Professional Identity Architecture, or PIA, is an early-stage framework for translating career evidence into demonstrated capability. It does not begin with the question:

Does this person match the expected resume pattern? It begins with: What capability is demonstrated by the evidence? That shift changes the work of identifying qualified candidates for their actual abilities and potential to succeed. Not candidates as potential placeholders. Not as disposable or interchangeable. Instead of treating a resume as the primary object, PIA treats the resume as a downstream projection of a deeper professional identity structure. The goal is not to inflate a person’s experience or produce marketing language. The goal is to preserve evidence, interpret it carefully, and make capability legible without flattening the person. A PIA-style professional identity output may organize experience around questions such as:

  • What was the purpose of the role or activity?
  • What did the person actually do?
  • What impact did that work have?
  • What did the person learn?
  • What capability appears across multiple contexts?
  • What evidence supports that interpretation?
  • What should not be inferred? This kind of structure is slower than a keyword scan. But it may be better suited to understanding people whose capability is real but not obvious in standard formats.

What OSI Is Trying to Measure

Organizational Systems Intelligence, or OSI, is the broader organizational mirror. If PIA asks how a person’s capability can be represented more accurately, OSI asks how organizations create, lose, misread, retain, or fail to deploy capability. In hiring, that means examining the organization’s own talent-intake system:

  • What does it filter out?
  • What does it reward?
  • What does it fail to measure?
  • What kinds of people become invisible?
  • What assumptions are embedded in job descriptions?
  • What evidence reaches human judgment?
  • What evidence is discarded before anyone sees it?
  • What does the organization count as a successful hiring process?

These questions matter because hiring is not only an HR function. It is a capability-recognition system. When that system works, organizations can identify people who can grow into the work, contribute to the mission, and strengthen the institution over time. When it fails, organizations may reject the very people who could have solved the problems they are hiring for.

Better Translation, Not a Faster Black Box

The future of hiring should not be a faster black box. It should be a better translation system. That does not mean abandoning technology. It means using technology differently. AI can help organize evidence. It can compare language across domains. It can surface transferable skills. It can help reviewers notice patterns that might otherwise be missed. It can support audit trails, consistency checks, and structured human review. AI should not become a substitute for judgment. The better path is not “human versus AI.” The better path is accountable interpretation:

  • use AI to expand the view, not narrow it;
  • audit what the system rejects;
  • preserve human review for ambiguous and high-impact decisions;
  • treat non-linear career evidence as signal, not noise;
  • document how conclusions are reached;
  • distinguish evidence from inference;
  • design hiring systems to filter in, not only filter out.

The hidden-workers research points toward this same basic direction: organizations should rethink hiring practices so they uncover overlooked talent pools rather than repeatedly exclude them.2 That is a practical design principle. The goal is not to lower standards. The goal is to stop confusing narrow legibility with actual qualification.

Conclusion

Hiring does not only have a talent problem. It has a translation problem. People are more complex than resumes. Capability develops across contexts. Evidence appears in different forms. Organizations need systems that can interpret that evidence responsibly. If hiring systems continue to optimize mainly for speed, surface polish, and pattern matching, they will keep missing people who can do the work. Some of those people will be hidden workers. Some will be career changers. Some will be veterans. Some will be caregivers. Some will be self-taught builders, informal leaders, systems repairers, operational stabilizers, or cross-domain translators. Some will simply be people whose real capability does not fit the expected template.

The next generation of hiring systems should help those people become legible without forcing them to become artificial. It should help organizations recognize capability before they lose it.

Sources

https://www.hbs.edu/managing-the-future-of-work/research/hidden-workers-untapped-talent

https://www.accenture.com/content/dam/accenture/final/a-com-migration/pdf/accenture-uncover-missed-talent-pools-improve-diversity.pdf

https://hbr.org/2026/06/ai-has-broken-hiring-heres-how-to-fix-it

https://verityai.co/blog/harvard-study-ai-rejects-qualified-candidates

https://ralionline.com/newsinsights/ai-adoption-is-accelerating-but-confidence-is-collapsing-the-more-workers-use-ai-the-less-they-trust-it-baby-boomers-show-a-35-drop/

  1. Harvard Business School, Managing the Future of Work, “Hidden Workers: Untapped Talent.” ↩

  2. Accenture and Harvard Business School Project on Managing the Future of Work, “Hidden Workers: Untapped Talent.” ↩ ↩2

  3. VerityAI, Sotiris Spyrou, “88% of Employers Admit: Our AI Rejects Qualified Candidates (Harvard Study).” Updated July 3, 2026. Secondary commentary/advisory interpretation of AI hiring risks and hidden-worker research. ↩

  4. RALI / Fortune syndication, “‘AI adoption is accelerating, but confidence is collapsing’: The more workers use AI, the less they trust it.” January 21, 2026. ↩

  5. Harvard Business Review, Shraddha Sunil and Mudit Saraf, “AI Has Broken Hiring. Here’s How to Fix It.” June 8, 2026. ↩