AI skills for college graduates: Exploring how instructors and employers prioritize AI skills differently

Synthesis: Ithaka S+R's AI Skills for College Graduates report surveys 500 US four-year-college instructors (April 2026) on how they prioritize 26 AI-related skills from the HiBob AI Skills Framework, and compares their ratings to a panel of US employers surveyed by HiBob. The headline finding is a systematic skills-prioritization gap: instructors and employers agree on the importance of only one skill (setting realistic expectations for AI-augmented work), with instructors favoring a critical, responsible-use orientation (limits of AI, attribution, human accountability) while employers favor workflow, automation, and human–AI teaming skills. Instructors also report their institutions lack both a consensus on what AI skills look like and an assessment framework for them — an infrastructure gap that parallels AI Literacy debates.

Background and the AI Skills Framework

The report is one of the first major efforts to vet a comprehensive AI-skills framework with college instructors. The framework, developed by HiBob (a multinational HR technology company) to understand which AI skills employers seek and pay a premium for, spans seven categories and 26 skills:

1. AI literacy — understanding human vs. AI capabilities, selecting appropriate tools, setting realistic expectations
2. Continuous learning orientation — iterating on AI tools, peer coaching
3. Prompting and input quality — clear actionable prompts, grounding in reliable sources, Multimodal AI prompting
4. Evaluating and improving output quality — proactive review, revising drafts for audiences, resisting sycophantic AI, bias & fairness awareness
5. AI safety, Ethics, and AI Governance — handling sensitive data, responsible use, human accountability, transparency & attribution
6. Workflow evaluation and redesign — mapping workflows, efficient human–AI handoffs, documenting decisions
7. Automation and technical integration — no/low-code automations, basic coding/API capabilities

The framework is designed to capture how a typical, non-technical employee uses AI tools at work — from foundational LLM chatbots to purpose-built systems.

Methodology

• Instructor survey (Ithaka S+R): 500 instructors from four-year nonprofit colleges responded (28,200 invited, April 13–24 2026). They rated each of the 26 skills on a seven-point importance scale, reported whether they taught each skill, and answered questions on institutional AI context.
• Employer data (HiBob): 1,200 employers surveyed Feb–March 2026; the report compares instructors against the US subset (200 employers).
• Respondent profile: mostly tenured/tenure-track (300) plus 122 adjunct/contingent; ~three-fifths at doctoral institutions and public universities; disciplines led by social sciences, natural sciences, and business.
• Limitations: the report notes potential early-response bias, exclusion of community colleges/vocational institutions, and that the "skill of doing things without AI" was a recurring instructor refrain.

Key Findings

A wide alignment gap, with one point of agreement

Across the 26 skills, instructors and employers agreed on the importance of only one: setting realistic expectations for AI-augmented work (an AI-literacy skill). In most other cases one group rated skills higher than the other.

Instructors prioritize responsible, critical use

Instructors ranked highest the skills reflecting responsible use of AI and navigating the limits of human and AI capability: uses AI responsibly, reinforces human accountability, transparency & attribution, proactively reviews output quality, and recognizes the limits of one's own expertise. One instructor noted that "understanding the limitations of AI and the fact that LLM cannot replace human intelligence is key." These priorities align with foundational academic values — attribution, review and revision, and information literacy.

Employers prioritize workflow, automation, and teaming

Employers valued skills related to workflows, automation, and efficient human–AI collaboration — the activities of workers rather than students. They also highly valued human-to-human and human-to-machine skills more than instructors, reflecting the team-oriented, efficiency-driven nature of workplaces compared to the individualistic evaluation common in academic settings.

The teaching gap is the strongest evidence of a skills gap

Only three of 26 skills are taught by half or more instructors. Two-thirds teach transparency & attribution (likely via academic integrity), and roughly half teach responsible use, proactive output review, and human accountability. But far fewer instructors teach the workflow-evaluation and technical-integration skills — precisely the categories where instructor–employer prioritization diverged most. The report calls this "the strongest evidence that there is an AI skills gap between higher education and employers": whole categories employers value are neither prioritized nor taught.

Institutional context: expectations exceed infrastructure

Instructors slightly disagreed that their institutions expect undergraduates to acquire moderate AI proficiency, and more strongly disagreed that institutions have a consensus on what AI skills look like or a shared framework to assess them. One instructor reported their "institution announced AI literacy as a top priority—and then did nothing since." These findings indicate institutions are in the early stages of organizing around AI-ready learning environments.

What this means for practice

• Instructors. Map your syllabus against the framework's 26 skills and add the categories almost nobody teaches — workflow evaluation and redesign, and automation and technical integration — since only three of the 26 skills are taught by half or more instructors.
• Instructors. Keep the critical-use skills you already prioritize (responsible use, transparency and attribution, proactive output review, human accountability) but tie them to the workplace tasks employers rate: documenting decisions, efficient human–AI handoffs, and career readiness, not only academic integrity.
• Instructors. State plainly where AI use is expected and where it is not, because the report found a recurring instructor position that doing things without AI is itself the key skill — a framing students need made explicit rather than left implicit.
• Administrators. Publish a shared institutional definition of AI skill outcomes and an assessment framework for them: instructors disagreed that their institutions have consensus on what AI skills look like or a shared way to measure whether students reach them.
• Administrators. Audit whole degree programs, not single courses, for AI-skill coverage — the instructor–employer prioritization gap is widest in exactly the categories that are least taught.

Limitations

• The instructor survey closed at a pre-determined 500 respondents out of 28,200 invited four-year, nonprofit college instructors, and the report states this may introduce early-response bias.
• The comparison group is much smaller and differently recruited: 200 US employers drawn from a 1,200-respondent HiBob panel surveyed in February–March 2026, compared against instructors surveyed April 13–24, 2026.
• Institutional characteristics and discipline were taken from an email list purchased from a marketing agency rather than reported by respondents, which the report says limits its reporting on institutional and disciplinary affiliation.
• The report excludes instructors from community colleges and vocational institutions, and its authors note possible positive bias because instructors with strong objections to AI may be less likely to respond.

Citation

Fried, M. (2026). AI skills for college graduates: Exploring how instructors and employers prioritize AI skills differently. Ithaka S+R.