Bridging the AI skills gap

A survey of business journal editors on graduate preparedness, AI adoption, and governance. Sole author.

Submitted to the American Journal of Undergraduate Research

Period
February to May 2026
Role
Sole author. Faculty mentor Dr Chong Huang, not a co-author
Approval
IRB2026-70, exempt, University of the Pacific
Tools
JMP Pro 17, Python, matplotlib, Power BI

170

complete responses

1,021

editorial board members in the frame

22

item instrument

The question

Business schools are adding AI to their curricula quickly. The people who referee the field’s published research are well placed to say whether graduates arrive prepared, and they are rarely asked. The study asks them directly, and separately asks what governance their own institutions have put in place.

Method

A 22-item instrument, fielded in Google Forms between February and March 2026 to a sampling frame of 1,021 editorial board members at ABDC A and A*-ranked business and management journals. IRB approval was obtained before any participant contact.

Analysis ran in JMP Pro 17 and Python. Descriptive statistics with distributional diagnostics, importance-performance analysis with paired comparison, internal-consistency estimation, exploratory Pearson and Spearman correlations with assumption checks, multiple linear regression, and descriptive k-means clustering. Figures in matplotlib, dashboards in Power BI.

Response and reliability
MeasureValue
Sampling frame1,021 editorial board members
Complete responses170
Response rate16.7%
Cronbach’s α, seven-item risk index.937
Spearman-Brown, three two-item composites.881 / .758 / .724
Minimum detectable correlation at N=170|r| ≥ 0.21, α=.05, power=.80
The survey instrument

PDF · 141 KB

The 22 items as they were shown to respondents, and a closing page from the manuscript’s methods that was not.

What it found

AI literacy was nominated a top-three curriculum gap by 66.5% of respondents. Only 10.6% called their institution’s AI guidelines clear and useful, which is the gap between adopting a technology and governing it.

Importance-performance analysis put all nine competencies in the high-importance, low-preparedness quadrant, with gaps ranging from 1.23 to 1.51. There was no competency the respondents rated as adequately covered.

A three-predictor regression explained 38.5% of the variance in perceived curriculum-employer alignment, an R² of .385.

Headline results
ResultValue
AI literacy a top-three curriculum gap66.5% of respondents
Institutional AI guidelines clear and useful10.6%
Competencies in the high-importance, low-preparedness quadrantAll nine
Importance-performance gap range1.23 to 1.51
Regression on curriculum-employer alignmentR² = .385, three predictors

Limitations

A 16.7% response rate on a specialist frame invites self-selection: editors who care about AI in the curriculum are more likely to answer a survey about it. The design is cross-sectional, so nothing here establishes direction.

The analysis stops where the sample stops supporting it. No factor analysis and no structural equation modelling were run, and the manuscript says so rather than leaving a reader to assume they were considered and passed.

The point

Reporting the analysis the sample supports, rather than the analysis that would sound most complete, is the whole of the method here.

Not published

The response data

Not published

Not published, and the reason is specific rather than cautious. 44 of the 84 combinations of discipline, years teaching and level are unique to a single respondent, the sampling frame is a published list of editorial boards, and every respondent wrote free text, some of it naming institutional detail. Suppressing columns does not fix a frame that anyone can obtain and cells that thin.