The hottest EdTech Substack posts right now

And their main takeaways
Category
Top Technology Topics

AI Can't Fix Student Engagement

After Babel • 1412 implied HN points • 04 Mar 26
🚌 Education EdTech
  1. Too many students are disengaged from school; only about one in three are highly engaged, and that lack of engagement undermines real learning and long-term outcomes.
  2. Engagement falls into four modes—Passenger, Achiever, Resister, and Explorer—with Explorer mode (curiosity plus agency) as the goal because it supports initiative, deep learning, and resilience.
  3. Parents and schools can move kids toward Explorer mode with concrete actions: model curiosity, give students choice and authentic projects, protect extracurriculars, manage tech, and use tools or workshops to make engagement visible and supported.

EdTech Is Borrowing Zuckerberg's Playbook

After Babel • 2125 implied HN points • 26 Feb 26
🚌 Education EdTech
  1. Saying “there is no evidence of harm” is often used to block action, but demanding product-specific causal trials is usually impractical or unethical, so converging correlational evidence should be taken seriously.
  2. Broad rollout of classroom technology — for example in Utah after 2014 — coincided with reversed gains in reading and math, suggesting widespread EdTech can correlate with stagnation or decline rather than clear improvement.
  3. When billions and millions of children are affected, the burden should be on proving clear, durable benefits before wide deployment; choosing restraint and investing in proven interventions avoids large opportunity costs.

Maybe Parents Are Just Misinvesting

Bet On It • 679 implied HN points • 02 Mar 26
🚌 Education EdTech
  1. After basic stability is ensured, most common parental investments (extra activities, enrichment, busywork) add almost no extra benefit. Only extreme rescue from neglect or truly exceptional, specialized effort produces large gains.
  2. Small, immediate rewards tied to demonstrated mastery (for example, paying for 100% scores on Khan Academy units) can drastically speed learning and cut costs compared with typical schooling. Short daily practice, immediate feedback, and deadline incentives produced multi-grade progress in the example given.
  3. Ordinary parents can get big returns by swapping low-value time and money sinks for simple high-ROI tactics like focused practice, frequent assessment, immediate feedback, and demand-side incentives. Basic literacy and numeracy can often be taught far faster and cheaper than commonly assumed.

AI and Academia

In My Tribe • 364 implied HN points • 24 Feb 26
🚌 Education EdTech
  1. New AI tools that can write, run, and manage code let individual researchers build scrapers, dashboards, and analysis pipelines far faster than before, creating a big gap between code-savvy users and ordinary users.
  2. Replacing junior researchers or coding projects with AI may be efficient for supervisors but it also destroys the hands-on training that turns students into skilled practitioners, so educators must find new ways to teach those capabilities.
  3. AI will make it much easier to churn out low-value papers, so the academic reward system needs redesigning to stop incentivizing quantity over meaningful research.

Pending Change in Higher Ed

In My Tribe • 470 implied HN points • 21 Feb 26
🚌 Education EdTech
  1. Parents are moving away from public K–12 toward private schools and homeschooling, which expands the pool of families willing to try alternative higher-education models like UATX.
  2. UATX expects a fast surge in enrollment that could quickly change campus culture and shows how new providers can exploit demographic and recruitment problems facing legacy universities.
  3. Colleges now face a governance choice about how much to embrace AI; going all in will reshape hiring, curriculum, and budgets but risks alienating faculty, while hesitating risks becoming irrelevant.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

Average is Over with AI

In My Tribe • 759 implied HN points • 01 Feb 26
🚌 Education EdTech
  1. The economy increasingly rewards people who can work well with AI, so those who complement machines will thrive while others doing automatable tasks lose opportunities.
  2. The Alpha model pairs AI-driven one-on-one learning in the morning with student-chosen project work in the afternoon, and it naturally selects for learners who can learn from AI and who push themselves to excel.
  3. Colleges and other institutions should move from lecturing and grading toward guiding and coaching, because conscientiousness and ambition plus the ability to use AI will determine who succeeds and will widen social and geographic divides.

Further Progress on AI Teaching Assistant

In My Tribe • 288 implied HN points • 15 Feb 26
🚌 Education EdTech
  1. The AI tutor tracks your skills and uses adaptive spaced repetition, showing items less often when you get them right and more often when you get them wrong, but it currently won't recognize you if you switch browsers.
  2. Building the tutor was fast with Claude, however the tool runs intensively so continuing development will require upgrading to a more expensive subscription.
  3. Universities suffer from too much professor autonomy and weak centralized leadership, which makes it hard to identify or reward instructors who teach AI-relevant skills and to reorganize the institution for the AI era.

How is teaching going?

In My Tribe • 364 implied HN points • 11 Feb 26
🚌 Education EdTech
  1. Seminar-style paper discussions don't fit less-prepared freshmen, so classes often turn into lectures; assigning core concepts and short written answers before class would establish a common baseline for discussion.
  2. AI can be used effectively for grading and feedback, as practice exams with AI grading matched instructor judgments; building an AI teaching assistant for intro courses is feasible but would take several months to a year of development.
  3. Student engagement is limited by competing commitments and time constraints, so active projects have produced mixed results and early morning classes reduce participation.

Tyler Cowen's AI Campus

In My Tribe • 531 implied HN points • 16 Jan 26
🚌 Education EdTech
  1. Higher education can be reshaped around AI: students pair with mentors while AI designs syllabi, lessons, and assessments, and a big part of current teaching should focus on learning how to work with AI during this fast-changing transition.
  2. Small, AI-built apps and free-form natural-language interfaces can replace clunky courseware, letting users ask plain questions like “When is my next paper due?” and get immediate answers, and these tools can be prototyped very quickly.
  3. Policy teaching should be comparative and skeptical: markets sometimes fail but governments also fail due to information limits and perverse incentives, so solutions should weigh Pigovian-style fixes against Coasean bargaining, community governance, and constitutional design.

Maybe a project?

In My Tribe • 151 implied HN points • 13 Feb 26
🚌 Education EdTech
  1. An AI teaching assistant could make freshman econ students fluent by using spaced repetition and testing them in new situations.
  2. A prototype demo for production possibility frontier exercises exists online, but it currently checks answers against hard-coded solutions rather than giving live AI corrections.
  3. The plan is to add real AI-driven feedback and a wider variety of examples so students get adaptive practice and become truly fluent.

Down the Rabbit Hole (On Purpose)

Gad’s Newsletter • 97 implied HN points • 23 Feb 26
🚌 Education EdTech
  1. Learning has three layers: domain knowledge (the what), methods (the how), and mindsets/wisdom (the why). Facts fade and methods need practice, but mindsets and wisdom endure and shape long-term judgment.
  2. AI will make domain knowledge and many techniques cheap and widely available, so educational time should be reinvested in mentorship, judgment, and mindset cultivation. AI can simulate scenarios to practice decision-making, but it can’t replace lived experience and human feedback.
  3. Durable learning requires spaced retrieval, varied practice, reflection, and apprenticeship, not just one-off content delivery. Classroom detours or 'rabbit holes' are often deliberate ways to build transferable judgment and help students learn when to trust a model and when to rely on intuition.

OpenAI's new "Study Mode" and the risks of flattery

Res Obscura • 3265 implied HN points • 31 Jul 25
🕹 Technology EdTech
  1. OpenAI's Study Mode is designed to help students learn by encouraging them to think for themselves instead of just getting answers. It uses techniques like asking questions and guiding discussions.
  2. While Study Mode could benefit some learners, it may also encourage flattery and make students feel good without necessarily promoting real learning. It's important for AI to challenge students, not just agree with them.
  3. Learning often works best in a group or engaging with others, rather than relying only on AI. Human interaction can provide necessary friction that helps students grow.

When to Graduate from College?

In My Tribe • 303 implied HN points • 28 Dec 25
🚌 Education EdTech
  1. AI may reduce the market value of formal credentials and shift hiring toward demonstrable skills, so the traditional diploma and a May graduation feel less climactic.
  2. Graduation should be flexible and based on readiness to work rather than a fixed number of credits; students might graduate when they start a job, join government, or launch a business.
  3. Colleges need to connect students with employers and adapt courses from day one, using a network-based model that emphasizes practical skills and connections over credentials.

Picturing my students

In My Tribe • 243 implied HN points • 26 Dec 25
🚌 Education EdTech
  1. The instructor has 34 students across three sections and used student photos plus an AI-built flash-card app to try to learn names, though the images had to be extracted manually first.
  2. AI coding tools are shifting from expecting professional toolchain knowledge to enabling "vibe-coding," letting amateurs create usable software without downloading or configuring complex developer environments.
  3. Students should learn vibe-coding, document their process with AI tools, and keep up with rapid AI coding progress so they aren’t handicapped entering organizations today.

Visions for AI

In My Tribe • 288 implied HN points • 12 Dec 25
🕹 Technology EdTech
  1. AI will eventually do most software engineering by taking English prompts to write and maintain business applications, making traditional developers unnecessary for routine work.
  2. Robots that understand and respond to human language will become much more useful, sparking a robotics boom and creating new roles for people who design practical uses for them.
  3. AI will automate many routine tasks in education and health care — personalized teaching software will handle factual instruction and AI tools could diagnose and treat — but political and institutional resistance means assisting human professionals will come first.

Fun, Not Fear: Rebuilding University Assessment for the Machine-Learning Era, & Going Beyond to Modes of Research-Analytical Practice

Brad DeLong's Grasping Reality • 222 implied HN points • 29 Dec 25
🚌 Education EdTech
  1. Use outside-the-class work plus short face-to-face interviews or check-ins to assess students, because oral exams stress-test real understanding and make grading fair even when students have powerful ML tools at hand.
  2. Teach students to use modern advanced machine-learning models as intellectual force multipliers by training them in the seven labors—survey, identify live issues, hone questions, research, analyze, store, and persuade—and by emphasizing provenance, triangulation, and small analytic scaffolds so tools accelerate thinking without replacing it.
  3. Recenter higher education on play and craft: make learning fun and practical by practicing prompting, debugging, oral explanation, and producing reusable artifacts, and budget the extra instructor time needed to do this well.

Two of Your Predictions for 2024 and Why They Are Probably Not Correct

Mathworlds • 1375 implied HN points • 17 Jan 24
🚌 Education EdTech
  1. Generative AI tools may not eliminate 90% of teachers' administrative tasks by 2024 according to a teacher survey.
  2. AI tutors evolving to become great is another prediction for 2024, but their widespread success remains uncertain.
  3. It's crucial for edtech developers to create tools that truly meet the practical needs of teachers and students, as indicated by survey results.

The EdTech Revolution Has Failed

After Babel • 2486 implied HN points • 12 Nov 24
🚌 Education EdTech
  1. Using too many digital devices in schools doesn't seem to help students learn better. In fact, it might be making their test scores worse.
  2. Students often get distracted by things like social media when they're supposed to be learning, which can hurt their ability to focus and remember what they've learned.
  3. Some schools are starting to move away from using devices in the classroom and focus more on traditional teaching methods, which might lead to better outcomes for students.

AI Education: Understanding the Hype

ChinaTalk • 504 implied HN points • 01 Aug 25
🚌 Education EdTech
  1. China's education system is very focused on exams, which means students spend a lot of time preparing for tests instead of using new technologies for learning.
  2. The government is trying to improve education by using AI and digital tools, especially to help rural schools catch up with urban ones.
  3. Without proper investment and access to technology, AI might not equalize educational opportunities, and families with more money may still find ways to get ahead.

CROSSPOST: DAVID DEMING: Using generative AI to learn is like Odysseus untying himself from the mast

Brad DeLong's Grasping Reality • 38 implied HN points • 20 Jan 26
🚌 Education EdTech
  1. Generative AI frees learners to explore ideas actively, like Odysseus untying himself from the mast. It lets people test, iterate, and learn by doing instead of just passively consuming information.
  2. Real progress in economic history comes when we stop treating the past as isolated anecdotes and instead treat it as a measurable, modelable system. Measuring, modeling, and running counterfactuals reveals how historical forces worked and why outcomes happened.
  3. Combining generative AI with system-focused methods promises new ways to analyze and experiment with historical and economic questions. That mix could let researchers and learners poke at counterfactuals and build richer, testable theories.

Who Needs a Professor?

UnfairNation by Ehsan Zaffar • 6 implied HN points • 24 Feb 26
🚌 Education EdTech
  1. AI can answer many questions, so traditional lectures and the professor-as-knowledge-delivery model are becoming obsolete. Teachers now need to change how they assess and teach.
  2. AI democratizes access to tutoring and expertise, giving students without elite resources personalized, always-available help.
  3. Humans still matter for mentoring: teachers can push students, model changing your mind, and evaluate real understanding in ways AI can't. That makes mentoring, judgment, and assessment design the new core roles for educators.

AI Does Not Fit the Shape of Schooling

Mathworlds • 550 implied HN points • 01 Jun 23
🚌 Education EdTech
  1. Schooling has a multidimensional shape with various purposes like cognitive development and social development.
  2. AI models need to align with the full visions for learning, beyond what AI can currently model well.
  3. In classroom settings, AI may have potential for teacher support and professional development, but may not fit within the primary vehicle for student learning.

Imperfect, but sometimes on the money

imperfect offerings • 179 implied HN points • 24 Nov 23
🕹 Technology EdTech
  1. Peter Thiel's Palantir has taken over the federated data service for the NHS, impacting data sharing opt-outs for patients and raising concerns about private interests in public health data.
  2. In the education sector, AI's influence, particularly in EdTech, raises issues around data governance, privacy regulations, and the challenge of regulating online platforms.
  3. AI's expansion into various sectors, including recruitment, poses challenges such as potential bias, pricing out of students, and the use of AI for assessments, leading to a possible 'AI-driven race to the middle' in hiring practices.

Accessing the Bonus Content in Your Curriculum

Mathworlds • 196 implied HN points • 18 May 23
🚌 Education EdTech
  1. Teachers find ways to access bonus content in the curriculum, going beyond what's expected.
  2. Good curriculum lets teachers offer bonus content to students through ingenuity and pedagogy.
  3. In engaging classrooms, students' ideas and thinking become the focus, essentially becoming the curriculum.

The Palindrome in 2026

The Palindrome • 3 implied HN points • 15 Feb 26
🚌 Education EdTech
  1. The focus this year is on deep technical teaching: finishing a hands-on machine learning book and publishing algorithm implementations, with more explainer and animated videos alongside written posts.
  2. Paid subscribers will get exclusive, intensive 4-6 hour workshops that compress course content into single-day sessions, starting with a mathematics of machine learning workshop and later a neural networks course.
  3. The operation is scaling by hiring regular contributing educators and investing in tooling like the open-source nb2wb to automate publishing Jupyter Notebooks and speed up production.

The Lifecycle of an Edtech Company - Part 1

system bashing • 176 implied HN points • 20 Jul 23
💼 Business EdTech
  1. Starting small and doing what you love can lead to building a strong community and following.
  2. As a business grows, the transition from a passionate creator to a structured company can bring challenges in managing relationships and revenue.
  3. Scaling a startup involves evolving into a company with defined roles, managing financial aspects, and adapting to market demands.

The "death zone" of venture capital ☠️

Friends of Parsnip • 78 implied HN points • 09 Jan 24
💼 Business EdTech
  1. Venture capital investments in edtech can be challenging due to limited successful outcomes
  2. Selling learning products is difficult, leading to poor venture returns in the edtech sector
  3. Combining learning with practical doing can create a successful business model in the food space

Life is imperfect

imperfect offerings • 79 implied HN points • 31 Aug 23
🕹 Technology EdTech
  1. Life is imperfect - The message shared is about life's imperfections and how it plays out in different settings, emphasizing the need to navigate through challenges.
  2. Criticism in edtech - Discussion on the critique of ed tech companies' practices, highlighting the need for addressing power imbalances and engaging with critical voices.
  3. Generative AI impact - Insights into how generative AI is affecting graduate employment, the restructuring of labor, and the broader impact on work routines and value.

I Finally Listened to You and I’m So Glad I Did

The Palindrome • 3 implied HN points • 15 Jan 26
🚌 Education EdTech
  1. A YouTube channel now hosts video versions of fan-favorite educational posts, with three "greatest hits" videos already uploaded.
  2. Subscribing is a quick, zero-cost way to support growth and help the channel reach more machine learning practitioners.
  3. The project aims to teach the fundamentals of math and machine learning clearly and steadily, avoiding hype and short-lived trends, with big plans for 2026.

Duolingo’s Scaling Journey: Education Revolution or Engagement Obsession?

Gad’s Newsletter • 26 implied HN points • 30 Jun 25
🚌 Education EdTech
  1. Duolingo uses fun games and personalized lessons to help people learn languages. This method keeps users engaged but raises questions about whether it's helping them learn deeply.
  2. The app has many users but faces challenges in getting people to pay for premium features. Most users stick to the free version, making it important for Duolingo to find new ways to encourage subscriptions.
  3. While Duolingo effectively attracts beginners, it struggles to keep advanced learners. The company needs to create better resources for those wanting to reach high levels of fluency.

Revisited and updated: How I reduced my marking time from 4 hours a week to 15 minutes

Eclecticism: Reflections on literature, writing and life • 3 implied HN points • 21 Dec 25
🚌 Education EdTech
  1. You can massively cut marking time by building praise and criticism comment banks and using spreadsheet formulas to randomize and concatenate feedback for quick copy‑paste delivery.
  2. Automatically generated comments can sound genuine and be useful, but they sometimes mismatch the student’s work and will require occasional clarification or manual edits.
  3. Spreadsheets are handy for tracking assignments and progress—functions like =randbetween() can generate scores quickly, though you may need to overwrite or adjust numbers and consider broader signs of progress like bravery.

Risks to teaching as work and teachers as workers

imperfect offerings • 39 implied HN points • 17 Jul 23
🚌 Education EdTech
  1. Teachers are vulnerable to automation and AI tools that could change the nature of their work and how it's valued.
  2. AI has the potential to impact various professions beyond teaching, such as journalism, acting, music, and art, through automation of tasks and production.
  3. The use of AI in different sectors, driven by profit motives, can lead to job insecurity and challenges to worker's rights across industries.

Underdog Founders - Charlotte left Meta to build Jackett

ChatGPT4 as a CEO and Underdog Founders • 19 implied HN points • 20 Jun 23
💼 Business EdTech
  1. Charlotte left a comfortable job at Meta to build Jackett, an AI-powered platform to empower teachers and save time on assessments.
  2. Jackett's mission is to unlock students' potential by automating educators' tasks through AI, helping teachers create custom problems quickly.
  3. Despite challenges, Charlotte secured funding, focused on the impact in India, and Jackett is now used by thousands of teachers, saving hundreds of hours monthly.

Maybe EdTech is the answer? (feat. Noam Gerstein, CEO and founder of Bina School)

State of the Future • 7 implied HN points • 12 Jun 25
🚌 Education EdTech
  1. Online education models like Bina School can lower costs by removing the need for physical buildings and administrative overhead that traditional schools have. This could push regular schools to rethink their business models.
  2. Instead of just preparing students for specific jobs, education should focus on helping them become adaptable, decision-making individuals. This means measuring success in new ways, beyond just test scores.
  3. The merging of education and publishing will change how content is delivered. Schools could become more responsive ecosystems, using real-time data to tailor learning experiences to students' needs.

The Data Science Education Consolidation

Sector 6 | The Newsletter of AIM • 39 implied HN points • 06 Jun 22
🚌 Education EdTech
  1. Edtech companies like BYJU'S and upGrad are buying smaller firms to strengthen their position in data science education. This shows a trend of growth and consolidation in the industry.
  2. Traditional training institutions like NIIT and Aptech are struggling to keep up with these changes. They seem to be losing relevance in the fast-paced education market.
  3. BYJU'S made a big impact last year by acquiring ten companies for $2.5 billion. This highlights the scale of investment happening in the education sector, particularly in data science.