GB710 Unit 3 Evidence Quality Appraisal Example

Four Sources, Four Answers: A Unit 3 Evidence Quality Appraisal on AI and Entry-Level Hiring

[Student Name]

Purdue University Global

GB710: Contemporary Challenges in Business

Unit 3 Evidence Quality Appraisal

[Instructor Name]

[Date]

The candidate and the staffing firm survey are composite and illustrative; the other sources and appraisal tools cited are real.

The Question

Is generative AI reducing hiring of entry-level professionals? The appraisal asks not which source is right, but how much weight each can bear and for what kind of claim. The distinction matters because the firm's partners have already cited all four, often in the same conversation, as if they were interchangeable. A survey of what hiring managers plan to do and a payroll study of what actually happened to young workers can both be accurate and still point in different directions, and treating them as equivalent is how a weak source ends up carrying a decision.

Method

Gray literature, meaning reports produced outside peer-reviewed journals, is appraised with the AACODS checklist (Tyndall, 2010), which asks about authority, accuracy, coverage, objectivity, date and significance. Research sources are appraised on design: how exposure to AI was measured, what comparison group was used, what outcome was measured and what else could explain the result. Each source receives an overall weight of low, moderate or high for the claim it is cited to support.

Source 1: A Staffing Firm's Hiring Manager Survey

A national staffing firm surveyed about 1,200 hiring managers in 2025 and reported that a majority planned to hire fewer entry-level staff because of AI. Authority: the firm has expertise in hiring but sells staffing services that benefit if employers believe their hiring needs are changing. Accuracy: the questionnaire was not published, and the sample was drawn from the firm's own client list. Coverage: mostly large employers in technology and finance. Objectivity: the report's conclusions align with the firm's marketing. Date: current. Significance: widely cited in trade press. Weight: low for claims about actual hiring; moderate for claims about what some hiring managers say they intend.

Source 2: The World Economic Forum's Employer Survey

The Future of Jobs Report surveys large employers worldwide about expected changes over five years (World Economic Forum, 2025). Authority: a well-known organization with a long-running survey. Accuracy: methods are described, though respondents are executives answering about expectations. Coverage: large multinational employers. Objectivity: no direct commercial interest in the answer, though the forum's audience is executives. Weight: moderate for claims about employer expectations; low for claims about current hiring.

Source 3: A Payroll-Based Working Paper

Brynjolfsson et al. (2025) use administrative payroll records covering millions of workers to compare employment trends by age and by occupational exposure to AI. Exposure measure: occupations are classified by how much of their work AI tools can perform, using existing exposure indexes. Comparison: younger workers are compared with older workers in the same occupations and with workers in less exposed occupations. Outcome: employment counts from payroll. Alternative explanations: the authors address interest rates and the post-2021 tech correction, but the paper remains observational and not yet refereed. Weight: high for the claim that early-career employment fell in exposed occupations; moderate for the claim that AI caused it.

Source 4: A Danish Study of Chatbot Adoption

Humlum and Vestergaard (2025) link surveys of Danish workers about chatbot use to administrative records of earnings and hours. Exposure measure: reported adoption by workers and employers in exposed occupations. Comparison: workers in firms that encouraged adoption against those that did not. Outcome: earnings and hours. Result: effects close to zero through 2024. Alternative explanations: Denmark's labor market institutions and the short time span may limit how far the result travels. Weight: high for the claim that adoption had small short-run effects in Denmark; low for United States accounting firms specifically.

How the Sources Travel Through the Press

The four sources are not equally visible. In the trade articles the candidate collected over three months, the staffing firm's survey was cited most often, usually in a headline, and the Danish study least often. The payroll paper was cited frequently but almost always for its most dramatic finding, without its caveats about causation. This pattern matters for the appraisal because the sources a reader encounters first are not the strongest ones. A partner who forms a view from trade coverage will have seen the weakest evidence most often and the strongest evidence least often, and in a simplified form. Part of the appraisal's purpose is to reverse that order, so that the firm's discussion starts from what was measured rather than from what was most repeated.

What a Better Source Would Look Like

None of the four sources examines accounting firms directly. A stronger source for the firm's question would compare firms that adopted AI tools at different times, measure their graduate hiring and the hours spent on first-year tasks before and after, and follow the graduates hired to see whether they progressed at the usual pace. Professional bodies and large accounting networks hold data that could support such a study. Until one exists, the firm must rely on evidence from adjacent settings and on its own records.

Disagreement Read as Information

The sources disagree partly because they measure different things. The two surveys measure expectations and intentions. The payroll paper measures employment changes among young workers by occupation. The Danish study measures earnings and hours among workers who adopted the tools. A decline in early-career employment, as in Source 3, could coexist with small effects on adopters, as in Source 4, if firms reduced hiring of new entrants while keeping existing staff. The disagreement therefore narrows the question rather than leaving it unanswerable.

Sources that measure different things are not really disagreeing; they are answering different questions about the same change.

What the Four Sources Support Together

Together the sources support three claims at different strengths. With high confidence: employment of young workers in AI-exposed occupations in the United States fell relative to comparison groups after 2022. With moderate confidence: many employers expect to hire fewer entry-level staff because of AI. With low confidence: AI has caused the decline in entry-level hiring. None supports a claim about mid-size accounting firms specifically, which is the gap the candidate's later units will address. For the firm's partners, the practical lesson is that the evidence justifies watching entry-level hiring closely, not cutting it on the assumption that AI has already made the work unnecessary. Rousseau et al. (2008) argue that management evidence should be weighed by its fit to the question, and none of the four fits the question the candidate's firm faces exactly.

References

Brynjolfsson, E., Chandar, B., & Chen, R. (2025). Canaries in the coal mine? Six facts about the recent employment effects of artificial intelligence [Working paper]. Stanford Digital Economy Lab.

Humlum, A., & Vestergaard, E. (2025). Large language models, small labor market effects (NBER Working Paper No. 33777). National Bureau of Economic Research.

Rousseau, D. M., Manning, J., & Denyer, D. (2008). Evidence in management and organizational science: Assembling the field's full weight of scientific knowledge through syntheses. Academy of Management Annals, 2(1), 475-515. https://doi.org/10.5465/19416520802211651

Tyndall, J. (2010). AACODS checklist. Flinders University.

World Economic Forum. (2025). The future of jobs report 2025.