πŸ› Campuses Doubled AI Spending. Is it Working?

What this means for educators + more

Welcome to Playground Post, a bi-weekly newsletter that keeps education innovators ahead of what's next.

This week's reality check: Institutionwide AI licenses doubled in one year, but only 23% of campus tech leaders say AI has improved teaching and learning. A study of tens of thousands of admissions essays found AI was supposed to level the playing field but lower-income students who use it are rejected more. And 39 states now require financial literacy.

Data Gem

FAFSA simplification is working. Nearly 10 million students are now eligible for Pell Grants, up 6% from the prior year. The biggest gains came from households earning $60,001 to $125,000, a newly reachable middle-income population.

Colleges Doubled AI Licenses in One Year

Institutionwide AI licenses at colleges more than doubled in one year, from 27% to 61% of institutions, according to the Inside Higher Ed/Hanover Research 2026 Survey of 130 campus chief technology officers.

Half say the return on investment is either unclear or fell below expectations.

Only 29% say AI investments met or exceeded their ROI expectations.

When asked where AI has delivered the most tangible value, 55% of CTOs cited employee productivity. 30% cited IT operations. 29% cited administrative efficiency.

Only 23% said AI has improved teaching and learning.

"This suggests that educators and students and faculty aren't finding AI to be all that useful," said Britt Paris, an associate professor at Rutgers and chair of the AAUP's Committee on Artificial Intelligence.

The spending isn't slowing down. 49% of CTOs say investing in generative AI is a high or essential priority, up from 34% in 2025. Agentic AI investment rose from 28% to 35%. Predictive AI from 24% to 32%.

What's driving the growth if the results aren't there?

Fear of falling behind. 

41% of CTOs ranked falling behind peer institutions in AI adoption among their top worries for 2030.

Brian Fleming, author of The Solution Trap, called it higher education's long-standing tendency to "innovate through imitation."

"One institution's AI strategy may not be right for the one down the street because that institution may be trying to solve a different problem," Fleming said. "Colleges need to slow down long enough to define the problem they're trying to solve."

Meanwhile, 55% of CTOs say skills and staff capacity are limiting AI's impact. 49% say the pace of change is unsustainable without new resources. And 59% say a critical cybersecurity breach is among their top worries for 2030.

For education innovators, the survey quantifies the gap between AI purchasing and AI impact. The product opportunity is in the evaluation layer: ROI measurement platforms tied to actual learning outcomes, not just productivity savings. AI governance dashboards that help institutions define the problem before buying the solution. And procurement analytics that benchmark AI spending against evidence of impact, so that the 41% buying from FOMO can see whether their investment is performing.

AI Was Supposed to Help Low-Income Students Get Into College. It's Doing the Opposite

Researchers at Cornell and Carnegie Mellon analyzed tens of thousands of admissions essays submitted to an unnamed selective institution over four years, spanning the period before and after generative AI tools became available.

Lower-income students, represented by fee-waiver recipients, were more likely to use AI in their essays.

They were also more likely to be rejected.

Even among students who used AI, lower-income applicants were rejected at higher rates than higher-income applicants.

The explanation isn't that AI is equally bad for everyone - it's that wealthy students use AI differently.

"High-income students have a lot of different resources. They have counselors, they have teachers, they have more support on top of ChatGPT," said lead author Jinsook Lee, a Ph.D. candidate at Cornell. Lower-income students "might only be able to use the free tier instead of the $200-per-month Claude, and the quality of the outcome of what free-tier ChatGPT gives us is really poor."

The study also found that essay language became significantly more homogeneous after AI's launch. 

The most convergence was concentrated among lower-income and rejected applicants.

AJ Alvero, a Cornell sociology professor and co-author, described the cost: "The essay is designed to give applicants an opportunity to highlight the idiosyncrasies of their life. If it's pushing all these applicants towards the same type of essay, the same template, students are inadvertently losing that opportunity."

The researchers framed their central finding in a single sentence: the shift has moved "from inequalities in access to inequalities in returns."

For education innovators, the finding reframes a bigger problem. As AI makes all essays sound the same, the personal essay becomes a weaker signal for admissions offices that depend on it to differentiate applicants. 

That creates demand for alternative evaluation formats: video interview platforms, project portfolio systems, and demonstrated-interest tracking tools that capture what an AI-homogenized essay no longer can.

39 States Now Require Financial Literacy

California and three other states just introduced financial literacy graduation requirements, bringing the total to 39 states.

The question is how to teach it without adding to an already packed school day.

Math and financial literacy educators say the answer is to combine the two subjects, starting as early as elementary school.

"Why not incorporate real-life contexts where students are using those concepts, through the lens of financial literacy?" said Latrenda Knighten, president of the National Council of Teachers of Mathematics. "They may plan a class party or special event. You've got a budget to work with to do that."

Or students could start a mock business selling friendship bracelets. Figuring out startup costs, pricing, profit margins, and repaying borrowed money.

"It makes sense to them," Knighten said.

The connection between math and money isn't new. It's ancient.

"People had financial issues they needed to address, so they developed tools," said Andrew Davidson, founder of Financial Life Cycle Education. "If you go back to Babylonian times, people started simple accounting to keep track of transactions. Counting things and having numbers comes from tracking inventory of loaves of bread."

Davidson's organization teaches financial literacy through four core concepts: 

  1. wealth (earnings and expenses),

  2. time (investing and borrowing), 

  3. risk (uncertainty and its instruments), 

  4. value (compensation for taking on or removing risk)

Each maps to existing math curriculum at every level. 

Elementary: counting and multiplication connect to time. Fractions and percentages connect to interest and discounts. 

High school: compounding interest connects to exponentiation and logarithms. Mortgage payments connect to series and sequences.

The approach doesn't require new lessons. It requires regearing existing ones.

For education innovators, 39 state mandates create a defined compliance market. 

Standards-aligned financial literacy curricula embedded in math instruction, not added as a separate course. Simulation platforms where students run mock businesses, manage budgets, and make financial decisions using real math. Teacher professional development that shows how to connect existing lessons to financial concepts. And assessment tools that verify students meet state requirements without adding testing time.

⚑️More Quick Hits

This week in education:

β€’ Denver mandates phonics and dyslexia screening after finding 67% vs 26% proficiency gap within the same district β€” Board visits revealed affluent schools using phonics-aligned instruction while lower-income schools in the same district did not, even after a districtwide curriculum rollout

β€’ Data centers near schools linked to lower math scores in first US study β€” Preliminary Brown University research found a statistically significant drop in third-grade math at schools within one mile of data centers, with sharper declines near multiple centers

β€’ Home-based child care providers losing half their enrollment as voucher shortages spread β€” A Bronx provider's enrollment fell to seven children by February, while the national median provider salary sits at roughly $13 an hour and a D.C. wage-supplement study showed $10-14K payments increased employment 7%

β€’ Only 45% of LA special education parents say their child feels safe at school β€” A survey of 342 families across nearly 40 districts found fewer than half said their child in special education often or always feels safe, highlighting gaps in case management and family communication

To stay up-to-date on all things education innovation, visit us at playgroundpost.com.

What did you think of today’s edition?

Login or Subscribe to participate in polls.