Hybrid Learning Environments

More Than Algorithms: How Hybrid Tutoring Is Rewiring Learning Equity Improving Student Learning with Hybrid Human-AI Tutoring is a three-site, quasi-experimental study led by Carnegie Mellon University, exploring the effectiveness of a hybrid tutoring model that combines AI-driven adaptive math software with human tutoring. Conducted in 3 urban, low-income U.S. middle schools, the intervention was designed to enhance learning for students historically underserved in math. The study evaluates outcomes among over 500 students—Black & Latinx—revealing that hybrid tutoring significantly increases student engagement and learning progress, particularly for students below grade level. At a fraction of the cost of traditional high-dosage tutoring, this model offers a scalable, equity-oriented solution to pandemic-era learning gaps [The cost of the hybrid human-AI tutoring intervention was reported as: Average cost per student: ~$700 USD/year] 5 Key Takeaways: 1. Hybrid Human-AI Tutoring Boosts Engagement and Progress: Students in the hybrid model showed statistically significant increases in time spent on task, lessons completed, and proficiency gains compared to students using math software alone. 2. Equity Gains: Hybrid Tutoring Reaches Students Who Need It Most: Students below grade level benefitted more from hybrid human-AI tutoring than their on-grade peers. AI-informed tutors were more likely to engage struggling students, even those who did not actively seek help. This suggests the model helps overcome systemic help-seeking disparities and redirects support toward the most underserved learners, advancing equity. 3. Teacher Support Helps Learning—But May Reinforce Inequities Without Guidance: While the presence of math teachers during EdTech sessions led to improved outcomes overall, these gains more benefited higher-achieving students. Teachers, without AI guidance, tended to respond more to students who actively asked for help. Hybrid tutoring systems equipped with dashboards can correct this imbalance by proactively identifying and prioritizing students in greater need. 4. Lower Tutor-to-Student Ratios Improve Impact: At one study site, reducing the tutor-to-student ratio from 1:8 to 1:4 significantly increased the number of learning modules completed per hour. This highlights how maintaining manageable group sizes is essential for maximizing personalized learning is key. 5. Quasi-Experimental Methods Offer Rapid, Useful Evidence—But Broader Validation is Needed: The study demonstrates how rapid-cycle quasi-experiments can provide timely and actionable insights into what works and for whom. Thomas, D. R., Lin, J., Gatz, E., Gurung, A., Gupta, S., Norberg, K., ... & Koedinger, K. R. (2024, March). Improving student learning with hybrid human-AI tutoring: A three-study quasi-experimental investigation. In Proceedings of the 14th Learning Analytics and Knowledge Conference (pp. 404-415). https://lnkd.in/eHyKtW7p

𝑵𝒐 𝒍𝒆𝒂𝒓𝒏𝒆𝒓 𝒔𝒉𝒐𝒖𝒍𝒅 𝒃𝒆 𝒍𝒆𝒇𝒕 𝒃𝒆𝒉𝒊𝒏𝒅 𝒊𝒏 𝒕𝒉𝒆 𝒏𝒂𝒎𝒆 𝒐𝒇 𝒆𝒇𝒇𝒊𝒄𝒊𝒆𝒏𝒄𝒚 𝒐𝒓 𝒔𝒄𝒂𝒍𝒂𝒃𝒊𝒍𝒊𝒕𝒚. This is one lesson that has stood out loud and clear to us at beVisioneers: The Mercedes-Benz Fellowship as we’ve scaled up our hybrid learning experiences over the past year. In the rush to scale, it’s easy to default to systems that prioritize the “efficient” over the “equitable.” 𝑹𝒆𝒇𝒍𝒆𝒄𝒕𝒊𝒏𝒈 𝒐𝒏 𝒂 𝒈𝒓𝒐𝒘𝒕𝒉 𝒂𝒏𝒅 𝒍𝒆𝒂𝒓𝒏𝒊𝒏𝒈 𝒚𝒆𝒂𝒓! Thinking how easy it could have been to be to be swept up by the allure of systems that work “for most” in a part-time fellowship program. But we were challenged by the presence of diverse learners who don’t fit into “most”? -Those with unstable or unreliable internet connections. -Those working full-time jobs. -Those caring for children or family members. -Those who are stretched too thin, not because of a lack of motivation but because life pulls them in too many directions. 𝐓𝐡𝐞𝐬𝐞 𝐚𝐫𝐞 𝐭𝐡𝐞 𝐥𝐞𝐚𝐫𝐧𝐞𝐫𝐬 𝐰𝐡𝐨 𝐡𝐚𝐯𝐞 𝐜𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞𝐝 𝐮𝐬 𝐭𝐨 𝐝𝐨 𝐛𝐞𝐭𝐭𝐞𝐫. 𝐀𝐧𝐝 𝐰𝐞'𝐫𝐞 𝐬𝐨 𝐠𝐫𝐚𝐭𝐞𝐟𝐮𝐥 𝐭𝐨 𝐭𝐡𝐞𝐦 𝐟𝐨𝐫 𝐭𝐞𝐚𝐜𝐡𝐢𝐧𝐠 𝐮𝐬 𝐬𝐨 𝐦𝐮𝐜𝐡! Paulo Freire (🐐) reminds us to honor the lived realities of learners. For us, it has meant rethinking and rebuilding learning systems from the ground up—not for the privileged few, but for EVERYONE by 🎟️ Designing low-bandwidth and easily downloadable and WhatsApp-able materials to ensure access for those with unstable internet. 🎟️ Building in asynchronous options 🎟️ Allowing flexible deadlines and alternative formats 🎟️ Leaning into community-driven learning so that no one learns in isolation. So grateful to Nihal Ahmed Maximiliano Cortes Sotomayor Joyce Zhang And Shruti Ryali for continually rising to the challenge and asking whose voices, needs, and realities are missing in the design process. I'm so proud of what we achieved in 2024 building a solid foundation for scale..and equity 🫶. To fellow educators and designers - How are you ensuring no one is left behind as you grow your reach? #pedagogyofkindness #pedagogyofflourishing

The term “blended learning” gained traction in the late 1990s, with one of its earliest documented uses appearing in a 1999 press release by the Interactive Learning Centers, which later became EPIC Learning A. However, the conceptual foundation was significantly shaped by Clayton Christensen and his colleagues at Harvard Business School. While Christensen is more widely known for coining “disruptive innovation,” his work in education especially in Disrupting Class (2008) helped frame blended learning as a student-centered model that reconciles standardization with personalized instruction. Implementing blended learning effectively in schools requires thoughtful planning, flexible structures, and a strong focus on student agency. Here are some research-backed strategies that can make a real impact: 🎯 Core Implementation Strategies • Start with a clear model: Choose from models like station rotation, flipped classroom, or flex model based on your students’ needs and school infrastructure. • Personalize learning paths: Use adaptive digital tools to tailor instruction, allowing students to progress at their own pace while teachers provide targeted support. • Blend modalities intentionally: Combine face-to-face instruction with online components that reinforce or extend learning not just digitize worksheets. • Train and support teachers: Provide professional development on digital tools, data analysis, and classroom management in hybrid environments. • Use real-time data: Leverage learning platforms to track student progress and adjust instruction dynamically. 🧠 Examples in Action • Station Rotation Model: In a study involving over 25,000 students, those using adaptive games like Prodigy in rotation stations met standardized test expectations nearly 12% more often than peers who didn’t. • Face-to-Face Driver Model: In Round Rock ISD, Texas, using interactive whiteboards in a blended format led to a 23% higher pass rate in Grade 5 math compared to traditional classrooms. 💡 Bonus Tips • Establish clear routines and expectations for both digital and in-person components. • Foster student autonomy by teaching digital citizenship and self-regulation. • Involve families to support learning beyond the classroom.

📢 New Report: Key Insights on Digital Learning in Canadian Post-Secondary Education I'm excited to share key findings from the 2024 Pan-Canadian Report on Digital Learning by Nicole Johnson at CDLRA/ACRFL ! This comprehensive report, based on surveys from Spring and Fall 2024, offers valuable insights into the current state and future trends of digital learning in Canadian post-secondary institutions. At D2L we were so excited to sponsor this research! Here are some of the key takeaways: Technology is Here to Stay: Respondents expect technology use in post-secondary education to continue increasing across all modalities, including hybrid and online learning. The most common technologies used in teaching and learning are learning management systems (LMS), online polling/quizzes, and video-based technologies. Hybrid Learning is Leading Growth: Of all learning modalities, hybrid learning is expected to grow the fastest. GenAI is on the Rise: There's a strong consensus that generative AI (GenAI) will become a normal part of post-secondary education within a few years, with many already using it for teaching activities and student learning. Future Outlook: While most respondents expect post-secondary education to be different in five years, there is an increased sense of pessimism compared to last year, stemming from concerns about funding, faculty support, and the sustainability of the current system. I encourage you to read the full report for a more in-depth understanding: https://lnkd.in/eqtDJWeV Let's discuss the implications of these findings for the future of education! #DigitalLearning #HigherEducation #EdTech #CanadianEducation #OnlineLearning #HybridLearning #GenAI #CDLRA #D2L

I talk to a lot of instructors about converting their instructor-led courses to online learning. I've developed a very particular way of doing it, but it is always a shock - instructor-led training is NOT a 1:1 transfer to eLearning. Both modes of training have very particular advantages and disadvantages. The chaos rendered on training by Covid was a great opportunity to showcase the advantages, but because it was overwhelmingly done poorly, there is a lot of prejudice against eLearning now. You can't replace a mechanic with a cook to fix a car and expect the same results at the end, but you wouldn't blame the cook, you'd blame the person who made the switch. eLearning is NOT a panacea and, quite honestly, has limited scope of utility. However, if paired into hybrid - eLearning coupled with instructor led online - it becomes powerful. A lot of what is delivered in class can be delivered in eLearning, then the instructor leads discussions, activities, or makes assignments. eLearning is also good for getting a lot of preparation out of the way for active classes: you can put policy and procedure into eLearning, then have a day of practicals and roleplay for reinforcement. Converting your instructor-led training should be performed by a #LearningStrategist or an #InstructionalDesigner who understands both sufficiently to mitigate the shortcomings of both, while magnifying the benefits. You need someone who has done both extensively to do the analysis and design, so they can mitigate the weaknesses of both and magnify the strengths, as well. #InstructionalDesign #LearningAndDevelopment #eLearningDevelopment

I'm working out loud today. A section in my book explores how hybrid AI-assisted workflows transform learning experience creation. This is my attempt at a visual. When I first started creating multimedia learning experiences in the 1990s, I used powerful but time-intensive tools like Adobe Premiere, Photoshop, and Audition. You know, create media in one tool and then import the content into an authoring tool like Storyline, Captivate, or Lectora. Many of those same workflows exist today but are rapidly changing with generative AI. With remote AI-driven tools, multimodal GenAI models, and headless content engines, we can automate, personalize, and scale learning experiences like never before. (BTW... the term "headless" is a software application that runs on a server without a graphic user interface. For instance, I can call a service like ElevenLabs for audio, which returns an MP3 file based on my request parameters.) ✅ Local Creation – Traditional tools give full creative control but require manual effort. ✅ Remote AI Content Engines – AI-generated text, video, audio, and images reduce production bottlenecks. ✅ AI Agents & Headless Workflows – AI agents coordinate content generation, integrating with LMS/LXP platforms via APIs. ✅ Avatar Content Pipelines – Text is transformed into AI-generated voices and videos, making training more engaging and scalable. ✅ The Future is Hybrid – Instructional designers will work alongside AI, leveraging automation without losing human creativity. Instead of spending months creating training content, what if an AI agent could dynamically generate learning materials on demand tailored to each learner? I’d love to hear your thoughts. How do you see AI impacting content creation in your L&D workflows?