←
Phil Hill26,418 posts
Phil Hill avatar

Phil Hill

@PhilOnEdTech

Ed tech consultant, blogger, family man

◷ Joined December 2008▧ 1,938 Photos and videos
Phil Hill avatar
Phil Hill@PhilOnEdTech·Nov 17

At MoodleMoot Europe in Barcelona, I heard a lot about digital sovereignty. The company I kept waiting to hear about was Mistral. That same week, the French AI company announced a model it had trained in its own European data centers using 3,800 Nvidia chips. I didn’t come across a session that took up that example, and I think we missed an interesting discussion. For a European university looking for an alternative to Anthropic or OpenAI, the interesting question is whether Mistral’s model could do the work. For a policymaker looking to reduce Europe’s dependence on American technology, those 3,800 Nvidia chips are also part of the story. I’d like to hear that conversation at the next EdTech conference. When a university says it wants more digital sovereignty, which decision does it want to be able to make for itself? My latest On EdTech post: onedtech.philhillaa.com/p/wh…

Post attachment
Phil Hill avatar
Phil Hill@PhilOnEdTech·Nov 11

Microsoft Classroom is worth watching. But let’s not overreact to a product announcement. Morgan and I have argued that the LMS product category is remarkably stable. The value of having a company handle security, updates, bug fixes, and support is increasing, not disappearing. That does not mean today’s LMS vendors are safe. Microsoft’s new standalone Classroom product explicitly targets higher ed as well as K–12. That distinguishes it from Google Classroom’s market focus and from the old approach of patching together Teams features into something resembling an LMS. Big Tech has the AI investment and campus relationships to change the competitive mix. But we’ve had the same Big Four LMS solutions for roughly 15 years, and those vendors are making their own AI investments. This is not an easy market to change. One disappointment: Microsoft’s initial interface looks awfully familiar. There’s that traditional LMS left column again. A way to make institutions comfortable, perhaps, but also a missed opportunity? And one question worth following: could Microsoft’s higher-ed ambitions push Google to become more aggressive with Classroom? The product category can remain stable while the companies competing in it change. onedtech.philhillaa.com/p/mi…

Post attachment
Phil Hill avatar
Phil Hill@PhilOnEdTech·Nov 3

“Undergraduate enrollment held steady” is technically an accurate description of the preliminary Fall 2026 numbers, but is also an inadequate guide to what colleges should be planning for. The National Student Clearinghouse reports undergraduate growth of 0.6%. My read is that 2026 marks the transition from the post-pandemic recovery toward renewed enrollment decline. Whether this fall ultimately finishes slightly positive or slightly negative is secondary to the direction of travel. Three observations matter here. The recovery is losing momentum while enrollment remains below its 2012 level. The demographic pressure from smaller entering cohorts is beginning to work through the system. And the age composition of enrollment growth deserves much more attention than the aggregate headline receives. Enrollment among students ages 17 or younger grew 3.9%, while enrollment among 18-year-olds fell 0.7%. Expanding dual enrollment has real benefits, but growth among high school students tells a different story from growth in the population colleges traditionally serve after high school. Institutions should keep improving access and retention. Those efforts matter, and outcomes will vary substantially by institution and geography. But finishing slightly above zero in Fall 2026 would offer little reassurance about the years ahead. My new On EdTech analysis: onedtech.philhillaa.com/p/th… #HigherEducation #Enrollment #DualEnrollment

Post attachment
Stephen Landry avatar
⟳ Phil Hill retweeted
Stephen Landry@landryst·Oct 29

HigherEd's EdTech Cybersecurity Reckoning - @PhilOnEdTech onedtech.philhillaa.com/p/ed… "The old internet promise that anyone could put a server online and keep it there has become much harder to sustain in the age of AI. ... The ability to put an application on the public internet carries an operating obligation that is becoming much harder to sustain through spare capacity, occasional upgrades, and the knowledge of one helpful colleague. I expect that pressure to drive a substantial shift toward managed services and shared infrastructure ... ."

Post attachment
Phil Hill avatar
Phil Hill@PhilOnEdTech·Oct 29

The Canvas breach was an early warning. PHP2Shell shows how much of higher education’s technology is exposed to the new economics of cyberattacks. AI is making it cheaper to find and adapt exploits. At the same time, colleges still have a long tail of public-facing applications run by departments, labs, and individual teams. When a critical vulnerability appears, someone has to know each application exists, identify the software beneath it, and act quickly. The response to PHP2Shell made that operating challenge visible. Some providers patched and restored services across their customers. Universities also restricted access to departmental sites while they worked out who could update them. I see this as a marker of a larger shift in EdTech: faster movement toward managed services and shared infrastructure, especially where institutions cannot fund continuous attention to every locally operated tool. AI may make useful software easier to create while making it harder to operate safely on the public internet. My new On EdTech analysis: onedtech.philhillaa.com/p/ed… #EdTech #HigherEd #Cybersecurity

Post attachment
Phil Hill avatar
Phil Hill@PhilOnEdTech·Oct 27

This is interesting news that I need to learn more about. @sethodell and his team at @KanahomaAgency . New investment partnership with Virent Capital to "support Kanahoma organically and acquisitively to become the next full service enrollment growth partner for institutions globally." And @chippaucek is joining the endeavor as co-chairman. A lot to unpack here and I suspect some interesting stories to emerge. linkedin.com/posts/chippauce…

Phil Hill avatar
Phil Hill@PhilOnEdTech·Oct 8

We need to stop calling every course chatbot an “AI tutor.” That fallacy is already leading to "research" (square quotes intended) that is impacting institutional decision-making. A new University of Maryland study reports lower average course grades and lower LMS participation in sections offered an AI study assistant. @morganmundum and I looked at what was actually tested (mostly based on Morgan's analysis, augmented by Stephen J. Aguilar): • Only about 15% of students offered access ever used the tool. • The default mode provided answers. Only two instructors ended the term with tutoring mode enabled. • Nearly 74% of student requests sought information, explanations, or solutions. Less than 1% sought feedback on students’ own attempts. Putting course documents behind a chatbot does not automatically create tutoring. Tutoring should identify what a student understands, diagnose misunderstandings, and adapt guidance accordingly. Likewise, putting a tool inside the LMS does not establish that instructors integrated it into their teaching. The lower grades and participation deserve attention. But the study does not establish that using an AI tutor caused students to learn less. At best, the study measures the effects of *offering this particular intervention with no scaffolding*, with limited adoption and an unresolved explanation for the declines. There is useful evidence here about a poorly-designed chatbot rollout. Presenting it as evidence about “AI tutoring” stretches that contribution into something the study barely tested. Availability is not adoption. Answer delivery is not automatically tutoring. Course grades and LMS activity do not tell the whole learning story. onedtech.philhillaa.com/p/an…

Phil Hill avatar
Phil Hill@PhilOnEdTech·Oct 1

High school students account for 9.3% of undergraduate enrollment at public four-year institutions—but 47% of their enrollment growth from 2022–23 to 2024–25. The composition of enrollment is important to understand. Building on CCRC’s new dual enrollment analysis, my latest On EdTech post looks at where the growth is coming from, by sector and institution. Across the institutions included in our analysis, high school dual enrollment students accounted for 38% of undergraduate growth. Some colleges are adding both high school students and other undergraduates. Others are growing overall because dual enrollment gains offset declines elsewhere. And this pattern extends well beyond community colleges. Rising headcounts alone don’t establish financial recovery or stability. onedtech.philhillaa.com/p/du… #HigherEducation #DualEnrollment

Post attachment
Phil Hill avatar
Phil Hill@PhilOnEdTech·Sep 21

I’ve held back on attributing the difficult job market for recent college graduates to AI. The tech hiring correction started before ChatGPT, and too many explanations skipped over that fact. We now know more. A new Census Bureau working paper offers some of the strongest evidence yet connecting college majors’ AI exposure to graduates’ actual employment and earnings. Its particular value: it follows graduates across industries, capturing setbacks that studies focused on particular occupations can miss. Graduates from the most AI-exposed majors experienced the largest deterioration. But these fields were also heavily exposed to the reversal of the pandemic tech hiring boom. The study cannot fully separate those effects. My interpretation: it is increasingly likely that AI is adding to the deterioration in early-career opportunities. How much remains unresolved. In my latest On EdTech post, I explain the findings and limitations in plain language, with recreated, interactive charts that let readers explore individual majors. Read “A Clearer Picture of AI and the College Job Market”: onedtech.philhillaa.com/p/a-…

Phil Hill avatar
Phil Hill@PhilOnEdTech·Sep 18

There’s a growing movement to ban or restrict AI in education. But once you get past the announcement, a basic question becomes surprisingly difficult: What exactly are you banning? AI is increasingly embedded throughout EdTech products. It can generate course materials, coach students, translate captions, remediate documents, summarize discussions, provide formative feedback, or evaluate student work. Those are very different activities, even when they sit inside the same product. In my latest On EdTech post, I look at the new restrictions in NYC, LA, Chicago, Berkeley, and Rutgers, then work through what they could actually mean for LMS and other EdTech products. I also look at the approaches emerging from MIT and the University of Sydney, which I find much more useful. They start with the educational activity and who gets to make the decision, rather than treating the mere presence of AI as the deciding factor. Because “we banned AI” is the easy part. Figuring out what that means in an LMS, courseware platform, Zoom meeting, or accessibility tool is where things get interesting. onedtech.philhillaa.com/p/wh…

Phil Hill avatar
Phil Hill@PhilOnEdTech·Sep 9

I think we can stop asking whether online education has “staying power.” The more interesting question is what the post-pandemic market actually looks like. The latest 12-month IPEDS data make the first part surprisingly clear. After years of steady growth, followed by the obvious Covid disruption, online participation appears to have found a new equilibrium. We now have four consecutive releases showing essentially the same pattern. This is starting to look less like the lingering effects of Covid and more like the emergent structure of US higher education. But the top line numbers hide quite a bit. Undergraduate and graduate students use online education very differently, despite apparently similar participation rates. And institutional online growth can mean actual enrollment growth—or merely offset losses elsewhere. I dug into all three, including why the newer 12-month data provide a better lens for this question than the Fall enrollment data getting attention this week. onedtech.philhillaa.com/p/on…

Post attachment
Phil Hill avatar
Phil Hill@PhilOnEdTech·Aug 24

IPEDS Fall Enrollment data gets most of the attention, but it has an important limitation for understanding online education: it is a snapshot taken in the fall. That increasingly misses how online education actually works, particularly with multiple starts, shorter terms, and working-adult students. The newly released 2024-25 IPEDS 12-month enrollment data gives us a different view. And there are some interesting results. Two-thirds of students took at least one online course during the year, considerably higher than the Fall Enrollment measure. The post-pandemic modality mix also looks remarkably stable at this point—not a continued return toward pre-pandemic patterns. And at the institutional level, WGU and Southern New Hampshire continue to operate at a scale that is hard to appreciate from the usual fall enrollment numbers. I pulled together the latest data, including trends by modality and sector and the largest institutions by online enrollment. onedtech.philhillaa.com/p/20…

Post attachment
Phil Hill avatar
Phil Hill@PhilOnEdTech·Aug 18

Risepoint announced this morning that it has acquired Keypath Education’s North American business. The headline is straightforward. The more interesting story is what Risepoint thinks it bought, and what the deal says about where the OPM market is going. I spoke with Risepoint CEO Fernando Bleichmar about the acquisition. His argument is that the old OPM model—built heavily around marketing, enrollment, and the bottom of the funnel—is stale. In this deal, Risepoint is making a specific bet on healthcare and particularly on clinical placements, a constraint that marketing can’t solve. There are also some interesting questions about what exactly Risepoint acquired, why Keypath sold now, and how much further this concentrates the OPM-supported nursing market. I cover the story in today's On EdTech post: onedtech.philhillaa.com/p/ri…

Phil Hill avatar
Phil Hill@PhilOnEdTech·Aug 7

I don’t often cross-post articles from other newsletters, but occasionally @morganmundum 's On Student Success covers a topic that deserves a broader audience. This is one of those cases. The post tackles a subtle but important shift in higher education. For years, growth mostly meant attracting more students into the system. Increasingly, however, growth means taking students from someone else. Particularly in the US and the UK. That may sound like semantics, but it changes how institutions should think about strategy. It also changes how we interpret stories about enrollment gains, partnerships, and “winning” in the market. It’s a useful framework for understanding why some colleges and universities are doing well despite a flat—or shrinking—overall market. Worth a read if you’re trying to make sense of where higher ed is headed. Original OSS post: onstudentsuccess.morganedtec… Cross-posted at OET: onedtech.philhillaa.com/p/wh…

Post attachment