My editors tell me I’m reaching my word limit, so let's jump straight to this week's news. First, we look at a report that challenges the story the media is telling about college. From there, we review new research about labeling first-generation students before wrapping up with some thoughts on Einstein, a new AI-powered learning companion. 

After reading today’s issue, share your thoughts about the Einstein AI tool in the comments! 

The College Story 

From What if the story we’re telling about college is wrong? | Lumina Foundation 

The latest Lumina Gallup report challenges the prevailing media narrative about higher education according to students currently enrolled. 

Our Thoughts  

I’ll be honest. I’ve been waiting for a report like this for a while. Anyone who follows the regular media narratives about higher education has watched an almost decade-long drumbeat of stories about fears of indoctrination, politicians declaring some degrees “worthless,” and campus crusades claiming there’s no free speech. Public confidence in higher education fell from 57% in 2015 to 36% in 2024, a genuinely alarming trend. But, if you read this blog regularly, you know that I’ve long thought that the broader cultural and political current is shaped more by media coverage and partisan messaging than by the lived experience of students.  

This report gives us a cleaner answer than we’ve had before. The students actually inside these institutions tell a markedly different story from the one dominating public discourse. Just 2% of all students, including 3% of Republicans, say they feel like they do not belong on campus because of their political views, with the majority saying all or most of their faculty members encourage students to share their views and support open discussion even on controversial topics. Roughly nine in ten students say their coursework is preparing them with career-relevant skills, and three-quarters of graduates say their degree has been critical or important to their career success. About nine in ten bachelor’s and associate degree students say the overall investment they are making is worth it. These are not the numbers you would expect if the system were as broken as its loudest critics suggest. 

However, the report is not all good news. Students overwhelmingly believe their degree is worth the investment, but they are increasingly clear that the pricing is neither fair nor sustainable. Only one in four students says four-year colleges charge fair prices. Just 31% of students rate private, not-for-profit four-year institutions as affordable. Even public flagship universities are only seen as affordable by about half of students. Community colleges remain the standout exception, rated as affordable by 93% of students and fair-priced by 53%. That gap is telling. 

What’s hard to ascertain from the report is whether students feel that way about the sticker price or the price they are paying after tuition discounts have been applied. While the net price that families pay has fallen for most, sticker prices continue to rise, so while economists may say that college is now more affordable, it may not feel like that to some families. Even after applying a $20,000 institutional grant to a $45,000 tuition bill, a $25,000 bill is still a lot.  

This is the issue I expect to dominate our sector over the next several years, and the structural forces behind it are not easy to address. Between Baumol’s Cost Disease, declines in state funding, and our deeply contradictory relationship with education in general, higher education will need to do more to demonstrate its worth while reducing costs. While students, the actual consumers of education, think the system is working, our cost trajectory will create real challenges that deserve serious attention. We have evidence that the foundation we provide sets up our graduates for meaningful careers and lives after college. Now, we just have to shape the media narrative to match that reality.  

First-Generation Students 

From Rethinking First-Generation Labels | Inside Higher Ed 

New research from the Common App looks at how we label first-generation students and how those labels may mask differences in student outcomes.  

Our Thoughts  

I’m a first-generation student. While my mother didn't even attend college, she made it absolutely clear from the time I was young that I (and my brothers) were going to college. Education was not a question in our household. It was an expectation, a value, and something she talked about with urgency. By today’s standards, some may look at my experience and wonder why she didn’t support alternative plans, but for my mother, she knew that getting an education was the path out of our lower SES lives.

By the technical definition used in most institutional systems, I check the first-generation box. But my experience navigating the idea of higher education was fundamentally different from that of a student whose family immigrated to this country and genuinely could not understand why their child would spend four years in school accumulating debt when they could be earning wages and contributing to the household right now. Both of us are "first-generation." The label does almost no work to explain the difference in what we each needed from an institution.

That is the core insight of this Common App report, and I think it is an important one worth considering. Based on parental education combinations and family structures, the researchers identified four clusters of students in need of varying levels of support. For example, the Striving group, which consists of students with a single parent in their life who did not complete college, completes degrees at rates 20 to 30 percentage points below their peers even after enrolling. That tells me that a meaningful subset of students we are currently grouping under a catch-all label are arriving on campus with a substantially different set of challenges, and we are not always equipped to see that clearly enough to respond to it. 

Parental education is not just a binary flag. It operates along a continuum that shapes a student's cultural capital, their comfort navigating bureaucratic systems, their sense of belonging, and their family's understanding of what college actually involves and demands. Luckily, we can begin to correct for this using existing data and a willingness to look more carefully at that data to shape how we classify students for support purposes. The binary first-gen label is not wrong. It just is not enough. 

New AI Tool Challenges LMS Usage 

From Agentic AI Can Complete Whole Courses for Students. Now What? | Inside Higher Ed 

Kathryn Palmer looks at a new AI agent, Einstein, that claims to be able to complete all coursework required in the Canvas learning management system. 

Our Thoughts  

Yikes. 

Okay, I won’t call it a day there, but I want to sit with that reaction for a moment, because I think it is the right instinct. An agentic AI tool that logs into Canvas, watches lectures, writes papers, participates in discussion boards, and submits homework autonomously is not a ChatGPT moment. It is something qualitatively different. OpenAI (or Anthropic or Google) put a powerful text generator in students' hands and asked them to make a choice about how to use it. Einstein removes the student from the loop almost entirely. The choice has already been made for them by the design of the product. 

Whether Paliwal's stated goal of sparking campus dialogue about credentialism was genuinely altruistic or a convenient post-hoc reframe once the backlash arrived is, I think, an open question. Launching a product explicitly marketed as a cheating tool, watching 124,000 people visit the site in three days, and then softening the language while keeping all the same functionality suggests the business logic and the pedagogical critique arrived at the same destination from very different starting points. But here is the thing: the underlying critique still has real force, and I do not want to dismiss it just because of how it was delivered. If a course is so transactional that a bot can complete it, maybe that’s a signal about how little deep learning is actually happening in that course. 

But here is where I part ways with the more dismissive version of Paliwal's argument, and where I think the "it's all just busywork" framing does real damage. Students are not great judges of what work is formative versus what work is filler, and that is not a criticism. It is a feature of being a learner. The problem sets that feel tedious are often the ones doing the cognitive heavy lifting. The reading that seems disconnected from the "real" material is often building the conceptual vocabulary that makes the real material legible later. Learning scientists have documented for decades that the kind of effortful, sometimes frustrating practice that students most want to avoid is often exactly the kind that produces durable knowledge and transferable skill. These are often referred to as "desirable difficulties," and the core insight is that ease of processing and depth of learning are frequently in tension. 

This matters enormously when we scale up. For example, you cannot do serious climate science without a working understanding of chemistry, physics, and systems dynamics. You cannot build the analytical judgment needed to solve hard problems in any domain by outsourcing the foundational work that builds that judgment. Yes, there are certainly some courses that feel transactional and credential-driven, but when students generalize anything that feels routine and boring as fair game for AI, they are robbing themselves of that foundation. The thinking they skip today is the capacity they will not have tomorrow. 

I know, it feels like I cover AI almost every week. I’m getting a little tired of it too. However, I hope the thread running through all of it is clear: the technology itself is not the story. How we choose to use it, govern it, and protect the things it cannot replace is the story. We are still in the early chapters. 

Update: As of this writing, the Einstein website referenced in the article is no longer active. Whether Companion took it down in response to the backlash or pivoted for other reasons is unclear. The capability it demonstrated, however, is not going away, which is exactly the point.