Before You Buy Another AI Tool: The Leadership Decision That Has to Come First

Institutional AI adoption in higher education jumped from 49 percent to 66 percent in 2025, and 88 percent of higher education professionals expect that growth to continue (Ellucian, 2026). Community college leaders are under real pressure to have answers. Boards are asking about it. Presidents are mentioning it in strategic plans. Vendors are calling with demos. And somewhere in the middle of all that noise, a budget decision is being made—often without a clear framework for what success looks like.

Most of those decisions are being made in the wrong order. Institutions are choosing tools before they have chosen a target. This pattern is visible in how leaders prioritize AI value: In a 2026 survey of more than 300 institutions, higher education administrators most frequently identified business and operations (68 percent); data and analytics (59 percent); and marketing, admissions, and enrollment (51 percent) as the areas where AI would deliver the greatest benefit—with student outcomes trailing well behind (Ellucian, 2026). Instead, the target—the specific student outcome AI is supposed to improve—is the leadership decision that should come first.

The Decision No One Is Making Explicitly

Every AI investment a community college makes falls into one of two categories: It improves either institutional operations or student outcomes. Both are legitimate goals, but they require different tools, implementation strategies, and accountability frameworks. Conflating these two categories produces a blind spot that no amount of technology spending can fix.

Operational AI, including faster workflows, automated advising queues, and better enrollment forecasting, is relatively easy to evaluate. Did the process get faster? Did costs go down? Did staff capacity increase? These metrics are familiar, and the data is accessible.

Student outcome AI is harder. If you deploy an AI-enabled advising tool, the right question is not whether advisors find it useful; it is whether students who interact with it persist at higher rates, complete credentials faster, or connect to employment more reliably. If you build an AI-enabled competency mapping system, the right question isn’t whether faculty can navigate it. It’s whether graduates are getting jobs.

Most institutions are measuring the first set of questions. Almost none are systematically measuring the second. The evidence base for AI’s impact on student academic achievement remains inconclusive in large part because institutions are not yet building it. Research confirms that the overall impact of generative AI on student outcomes lacks consensus, with effect sizes reflecting substantial variability across contextual conditions and methodological approaches and the overall impact remaining inconclusive without controlling for baseline differences across studies (Chen & Cheung, 2025). That is not a technology problem. It is a leadership problem.

The gap persists, in part, because leaders have not yet made the underlying decision explicit. Institutions moving from AI experimentation to meaningful student impact consistently share one characteristic: a deliberate choice about which outcomes they are building for before selecting tools (Achieving the Dream, 2025).

A Framework for the Decision

Before approving any significant AI investment, community college leaders should be able to answer three questions.

• What specific student outcome is this investment designed to improve? Leaders should not cite a general goal, such as “enhance the student experience” or “improve completion.” Rather, a specific measurable outcome with a baseline and a time horizon should be identified. If the answer is, “we want to stay current with technology” or “our peer institutions are doing this,” that’s a reason for keeping pace, which may be valid. However, it is not a student outcome goal.
• Who are the students this investment is designed to serve, and are they the students with the most to lose if it fails? Community colleges serve first-generation, adult, and workforce program students who have the most to gain from AI-enabled pathways, and the least margin for error when those pathways don't deliver. An AI investment aimed primarily at students who are already succeeding is a different investment than one aimed at students who are not. Both may be defensible, but the choice should be explicit.
• How will we know if it's working, and for whom? Establish a review cycle, set a baseline, and commit to examining disaggregated outcome data—not just overall metrics—before the contract renewal conversation. Without this, every AI investment becomes self-justifying.

What This Looks Like in Practice

Cuyahoga Community College’s ASCEND initiative, supported by the Ohio Department of Higher Education, was built on exactly this sequence. The college started with a student outcome goal: Give students in nursing, STEM, and business programs a portable, documented record of their competencies connected to labor market outcomes, so the connection between their learning and a career is visible and credible before they graduate. That goal determined the technology we needed, not the other way around. It required alignment across academic affairs, workforce development, career services, and employer partners before a single software purchase was made. It also gave us a clear accountability framework: building student records that employers can evaluate and ensuring that those records connect to employment outcomes. The tools came last. The leadership decision—about which students we were building for and what success looks like—came first.

The Conversation Worth Having

AI is not going away, and the pressure to invest in it is only going to increase. The question for community college leaders is not whether to invest; it is whether to invest with intention, and with a clear picture of which students will benefit and how outcomes will be tracked before a contract is signed.

The institutions that get this right may not be the ones with the most sophisticated technology. They will be the ones that made a clear leadership decision about which students they were building for, aimed their AI investment at that target, and held themselves accountable to outcomes that mattered.

That decision is available to every community college leader right now. It doesn’t require a bigger budget or a better vendor. It requires asking—and answering—the right questions before the purchase order is signed.

References

Achieving the Dream. (2025). Creating the AI-enabled community college: A road map for using generative ai to accelerate student success. https://achievingthedream.org/wp-content/uploads/2025/07/ATD-AI-Task-Force-Report-Final-7-21-25.pdf

Chen, S., & Cheung, A. C. K. (2025). Effect of generative artificial intelligence on university students’ learning outcomes: A systematic review and meta-analysis. Educational Research Review, 49, 100737. https://www.sciencedirect.com/science/article/pii/S1747938X25000740

Ellucian. (2026, March 4). Ellucian's 3rd annual higher education AI survey signals shift from individual AI use to institutional strategy, data privacy still the top barrier. PR Newswire. https://www.prnewswire.com/news-releases/ellucians-3rd-annual-higher-education-ai-survey-signals-shift-from-individual-ai-use-to-institutional-strategy-data-privacy-still-the-top-barrier-302704074.html

Stephen Griffin, M.B.A., Psy.D., is Chief Learning Officer, Workforce Innovation, and Vice President, Skills-Based Education and Career Pathways, at Cuyahoga Community College in Cleveland, Ohio. He is also a Fulbright Specialist (2024-2027).

Opinions expressed in Leadership Abstracts are those of the author(s) and do not necessarily reflect those of the League for Innovation in the Community College.