The research examines whether growing AI use is accompanied by judgment students can demonstrate.
Since the 2022-23 generative-AI inflection point, the research program has followed business students as AI moved from novelty to routine academic and professional use. The central questions have also changed. Current studies focus on whether students can verify important claims, explain how AI shaped their work, set appropriate human-approval boundaries, and make defensible decisions when the technology becomes more capable.
Adoption data tells us who is using AI. Learning evidence has to tell us what they can do with it.
Business schools increasingly know that students are using generative AI. Usage frequency, familiarity, and attitudes remain useful signals, but they cannot establish whether a student can supervise an AI-assisted workflow or recognize when an output deserves more scrutiny.
The research therefore looks for transfer. It compares stated beliefs with scenario performance, disclosure behavior, verification routines, oversight decisions, and open-ended explanations. Those comparisons reveal where confidence and good intentions carry into practice and where a capability still needs deliberate instruction and assessment.
This same distinction shapes Navigate AI's institutional work. A school may have meaningful AI activity while still needing stronger evidence of what students can demonstrate. The research helps identify the human routines that deserve direct attention.
Students strongly endorse human review, yet many are less ready to supervise AI in a concrete situation.
In the Spring 2026 Future Leaders sample, 90.8% said they knew when human review was needed. The stricter ready-supervisor screen was met by 47.1%. The 43.7-point difference directs attention toward applied oversight: where approval should occur, what a person should verify, and when AI should be prevented from acting without human review.
Knowing the principle and performing the practice often produce different signals.
The strongest findings come from moments when self-report can be compared with behavior, scenario judgment, or written explanation. Those gaps are especially useful for curriculum and assessment because they show where additional practice can be designed.
The Disclosure Translation Gap
In the Spring 2026 sample, 67.9% said they knew where and how to acknowledge AI use. When students had to write the disclosure, 31.2% met a minimum-complete standard and 14.3% produced a stronger explanation that addressed process, verification, and responsibility.
The result suggests that disclosure works better as a practiced professional communication skill than as a policy reminder. Students need examples, repeated use, and feedback on what a meaningful disclosure communicates.
Course exposure changes the conditions for fluency faster than judgment performance
Students with AI-focused coursework showed substantially higher weekly use, disclosure readiness, disclosure behavior, institutional support, and multi-step use. Differences on scenario-based judgment were much smaller.
The pattern matters for program design. Exposure can increase practice and normalize responsible-use routines, while calibrated judgment still requires assignments that ask students to diagnose, verify, revise, and explain.
Self-reported readiness provides an incomplete picture of demonstrated judgment.
The Future Leaders analysis and a separate study of emerging marketing professionals both point toward a calibration problem. Students can feel comfortable with AI while important verification skills remain uneven.
80.8% say they verify important AI claims.
Another 86.7% say they understand when AI should not be trusted. Yet 53.3% missed at least one problem in scenario-based tasks. A smaller high-self-report/lower-performance group represented 15.8% of the analytic sample.
Perceived skill was essentially unrelated to demonstrated proficiency in a separate N=308 study.
Self-reported AI skill and demonstrated proficiency had a near-zero relationship, r = -0.02. The finding supports a simple educational implication: confidence should be supplemented with observable evidence of verification, judgment, and responsible workflow design.
Assessment should make the human work visible.
Scenario diagnosis, verification logs, process explanations, approval rules, and defense prompts provide stronger evidence than asking students whether they feel prepared. These practices also give faculty something concrete to improve when the evidence reveals a weakness.
Students are willing to build with AI while placing boundaries around what they share.
The Reluctant Architect study, based on N=308 emerging marketing professionals, examines how trust, usefulness, perceived skill, and privacy concerns shape AI use. The findings describe a user who wants the capability of AI without giving the system unrestricted access to personal data.
4.1
Mean intended use on a five-point scale.
3.2
Lower willingness to provide personal information to AI-enabled systems.
β .51
Trust was a substantially stronger predictor of intended use than perceived usefulness, β .23.
The study calls this pattern the AI Trust Paradox. For educators, the finding expands the readiness conversation beyond technical skill. Students also need practice deciding when reliance is warranted, what information should remain outside a system, and how to retain accountability when AI becomes embedded in professional workflows.
Multi-step AI use is already common enough to require explicit approval boundaries.
In the Spring 2026 Future Leaders sample, 55.0% reported weekly or more frequent multi-step or agentic AI use. Among weekly users, 53.0% did not consistently select the strictest human-approval expectation before an AI system takes external action.
Agentic systems change the nature of AI literacy. A student who delegates several steps to an AI system has to decide what the system may do independently, where a person must intervene, what should be logged, and how a mistake would be detected before it creates an external consequence.
That moves the educational problem from prompting toward supervision. Courses can respond with explicit go/no-go rules, risk tiers, audit trails, verification checkpoints, and assignments that require students to defend where they kept human approval in the workflow.
Prepared to prompt. Developing the ability to supervise.
This stream is being developed as a conference-first research direction because terminology and practice around agentic AI are still evolving quickly. The underlying educational question is already actionable: what supervision routines should students practice before workplace systems act on their behalf?
Three years of repeated cross-sectional work trace a shift in the student AI conversation.
The longitudinal research agenda began with the disruption created by generative AI and has continued through successive student cohorts. The studies are reported separately rather than combined into one cumulative sample because each year answers a somewhat different question.
Fear and policy
Early work focused on integrity, uncertainty, and students' need for clearer institutional expectations as generative AI entered academic work.
Belief and adoption
Later studies examined the gap between strong career belief in AI and actual use, including the roles of ethical readiness, trust, privacy, and policy ambiguity.
Utility and depth
Recent work finds AI embedded in routine creation while analytical, verification-heavy, and multi-step uses remain more uneven.
Transfer and oversight
The current agenda focuses on demonstrated judgment, verification, disclosure, human approval, and the evidence needed to show those capabilities.
The sequence describes the evolution of the research questions, not a causal stage model for every student or institution.
Several studies examine different parts of Human+AI readiness.
Each stream has its own sample, design, and status. Together they build a fuller picture of how AI use interacts with judgment, trust, privacy, communication, assessment, and human-centered professional practice.
| Research stream | Central question | Contribution to practice |
|---|---|---|
| Future Leaders AI Readiness 2026 | Do verification, disclosure, and supervision principles transfer into applied performance? | Supports scenario-based practice, direct evidence, and calibration between confidence and demonstrated capability. |
| The Reluctant Architect | How do trust, usefulness, privacy, and perceived skill shape AI use? | Highlights calibrated reliance, data boundaries, and the Confidence and AI Trust Paradoxes. |
| From Novelty to Utility | How is student AI readiness changing across successive cohorts? | Tracks the movement from disruption and adoption toward depth, judgment, and professional utility. |
| Bridging the AI Readiness Gap | What helps students move from belief in AI's importance to active use? | Shows the roles of ethical readiness, trust, privacy, and clearer educational support. |
| Beyond AI Disclosure | How can AI governance become part of the learning design? | Develops the Disclose, Document, Verify, Defend practices used in assessment and course design. |
| The Seven Anchors | Which human capabilities remain central as marketing work becomes more AI-enabled? | Connects AI fluency with understanding needs, judgment, creativity, trust, purpose, and other human-centered capabilities. |
The studies inform what Navigate AI looks for in courses, assessments, and student evidence.
Research findings become useful when they change what a school asks students to demonstrate and what faculty look for in the evidence.
The research informs the questions and practices used by Navigate AI. Validation of the institutional Diagnostic and evidence of client outcomes are evaluated and reported separately.
The work moves between scholarship, teaching, and institutional practice.
The research program includes conference papers, journal manuscripts, professional reports, faculty-development sessions, and applied frameworks. The work has received Best Paper recognition at the Marketing Educators' Association and has been presented across successive years at the Academy of Marketing Science, alongside university and professional presentations on AI, marketing education, and human-centered practice.
The goal is practical as well as scholarly. Findings that survive analysis should improve the questions faculty ask, the assignments students complete, and the evidence business schools use to judge whether AI integration is producing the intended learning.
Use evidence to move the AI conversation from activity to demonstrated capability.
Explore the institutional Diagnostic or discuss a course, assessment, or curriculum question where stronger evidence would help.