Last year I wrote an upbeat tour of the education research that excited me. This year I want to start somewhere less comfortable, because 2026 has earned it. Artificial intelligence didn’t just give students a new tool. It held up a mirror to the entire system, and forced a question most of us in education have spent decades avoiding:

Have we been teaching students to learn, or just to produce output?

As the founder of ConnectPrep, I sit with students and parents every week. As a Washington University in St. Louis alumnus, I watch how a serious research institution responds when the ground shifts. From both seats, 2026 looks like a reckoning, and reckonings are where the real opportunities hide. So here are five hard truths I believe about education right now. Each one is a little uncomfortable, and each one is backed by what’s actually happening. I’d love your pushback in the comments.

Hard Truth #1

The 4.0 GPA is the most overrated credential of 2026.

For a generation, we told students that the transcript was the prize. In 2026 that promise is quietly collapsing. When employers were shown profiles of recent graduates and asked who they’d hire, the candidate with the 4.0 GPA finished last, behind every candidate with an internship or real industry experience. Meanwhile, generative AI has exposed a problem grades hid for years: a polished assignment no longer proves a student understood anything. As one widely-shared Inside Higher Ed essay put it, AI revealed where learning was “thin to begin with.”

This isn’t an argument against rigor. It’s an argument that the signal we optimized for, the clean transcript, has decoupled from the thing we actually care about: can this person think, build, and adapt? Students who treat the GPA as the finish line are training for a race that already ended.

Hard Truth #2

AI broke the college essay, and “test-optional” along with it.

The 2025–2026 admissions cycle brought some of the biggest changes to essay evaluation in a generation. Schools dropped numeric essay scores, added prompts asking applicants to reflect on their own relationship with AI, and began warning that AI-generated text could cost a student their spot. Roughly 40% of four-year colleges now run AI-detection tools, and a growing number use AI to triage applications before a human reads them.

Here’s the contrarian part. Test-optional was sold as the great equalizer. But when the essay, the most “holistic” part of the file, can be generated in nine seconds, the things that still differentiate a candidate quietly shift back toward the measurable. It’s no accident that elite schools are drifting back toward requiring the SAT and ACT, citing predictive validity. AI didn’t democratize admissions. It made authenticity the new scarce asset, and authenticity can’t be prompted.

In an AI world, the most valuable thing a 17-year-old can show an admissions office is a mind that’s unmistakably their own.

Hard Truth #3

The real skills gap isn’t technical. We automated the thinking we were supposed to teach.

Everyone is talking about the “AI skills gap,” and it’s real. Only about half of graduates feel they have the AI skills employers want. But focus only on tool fluency and you’ll miss the deeper crisis. When executives rank what they actually need, critical thinking comes first, ahead of communication, collaboration, and technical ability. And here’s the warning shot: Gartner predicts that the erosion of critical-thinking skills from over-reliance on generative AI will push half of organizations to require “AI-free” skills assessments by 2026.

Sit with that. We’re so worried students can’t use AI that we’ve under-noticed the opposite danger: that they’ll outsource their judgment before they ever develop it. The student who lets AI do the first draft of thinking, not just typing, graduates fluent in prompts and bankrupt in reasoning. The durable skill of the decade isn’t using AI. It’s knowing when not to.

Hard Truth #4

Most edtech in 2026 is still solving the wrong problem.

Walk any edtech expo floor and you’ll see dashboards measuring engagement, time-on-task, and streaks. Beautiful metrics. Mostly the wrong ones. AI forced a brutal clarification of what education is for: if a student can generate acceptable work instantly, the value of a course now rests entirely on whether they can explain, critique, and adapt that work in a new context. Engagement was never the goal. Understanding was.

The tools I’m betting on in 2026 don’t ask “how do we keep students clicking?” They ask “how do we make students defend their reasoning out loud?” That’s a much harder product to build and a much harder metric to fake, which is exactly why it matters. The next great edtech company won’t win by adding AI. It’ll win by using AI to make thinking visible again.

Hard Truth #5 (and the hopeful one)

WashU’s +AI bet is the contrarian-correct response, and it’s why I’m optimistic.

Here’s where I get to be hopeful, because my alma mater is doing the thing I wish more institutions had the nerve to do: refusing to choose between embracing AI and protecting what makes learning human. This spring, WashU launched a university-wide initiative simply called +AI, led by Provost Mark D. West. The framing is the part I love: it explicitly centers student formation alongside AI adoption, instead of treating the two as enemies.

And it’s not a press release. It’s shipping. Through the Digital Intelligence & Innovation (DI2) Accelerator, WashU is standing up an AI Curriculum Corps this summer to help faculty redesign courses around generative AI’s strengths rather than bolt it on. This fall, the McKelvey School of Engineering introduces a new minor in computational AI, and seed grants are funding interdisciplinary pilots that pair machine learning with real pedagogical questions. That combination of bold adoption and fierce protection of judgment is the exact posture I’ve been arguing for in this whole piece.

The institutions that win the next decade won’t be the ones that adopt AI fastest. They’ll be the ones that stay clearest about what it’s for.

That’s the thread connecting all five truths. The 4.0, the essay, the skills gap, the dashboards, the campus strategy: every one of them is really a question about whether we’ll keep optimizing for the proxy or recommit to the real thing. WashU is choosing the real thing. So is ConnectPrep.

What this means for students, parents, and the rest of us

If you’re a student: stop collecting credentials and start collecting evidence: things you’ve built, defended, and can talk about without notes. If you’re a parent: ask your kid to explain their homework back to you, not just finish it. If you’re a fellow founder or educator: build for understanding, not engagement. And if you’re hiring: look past the transcript for the mind behind it.

At ConnectPrep, this is the shift we’re building every program around: helping students develop the judgment, voice, and authentic story that no model can generate for them. That’s also the heart of our career coaching work: turning capable people into candidates who can prove they think.