INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES KNOWLEDGE PACK Technologies for Personalized and Adaptive Learning EXPLORE INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES KNOWLEDGE PACK TECHNOLOGIES FOR PERSONALIZED AND ADAPTIVE LEARNING © 2022 International Bank for Reconstruction and Development / The World Bank INDEX 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org INTRODUCTION Problem statement License: Creative Commons Attribution CC BY 3.0 IGO WHO WHY Possibilities Evidence This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. This report was also supported with funding from the Deployment process Global Partnership for Education. Structure of solutions The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, What has been done WHAT HOW Costs and budgeting denominations, and other information shown on any map in this work do not imply any judgment on the in other countries Monitoring and part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such evaluation boundaries. Some references may appear in this Knowledge Pack to Logos, Products, Brands or Trademarks belonging to CONCLUSION others not affiliated with the World Bank. They belong to their respective owners/ holders and are used for illustrative purposes only and do not imply any affiliation with or endorsement by them. The World Bank does TO GO FURTHER not endorse, prefer or recommend any of these products. ANNEXES Rights and Permissions Glossary The material in this work is subject to copyright. Because The World Bank encourages dissemination of its References knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. FAQ Please cite the work as follows: EdTech team. 2022. Knowledge Pack: Technologies for personnalized and adaptive learning. Washington, D.C.: World Bank Group. Acknowledgment: KP developed under the guidance of Iñaki Sánchez Ciarrusta and Cristóbal Cobo from the EdTech Team. Much appreciation to Omar Arias and Jaime Saavedra for their Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank overall support. Also we want to thank Juan Baron, Diego Angel Urdinola, Tom Kaye and the rest of Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; colleagues who provided comments to enrich these resources. e-mail: pubrights@worldbank.org Design : Alejandro Scaff, Sarah Kleinmann 2 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Introduction Overall, the common questions addressed in this KP are the following: What is a KP? About this KP GENERAL QUESTIONS Knowledge Packages (KPs) The objective of this KP is to • Why technologies for personalize and adaptive learning can be part are short, pragmatic guides provide sufficient knowledge of the solutions? on individual topics within on personalize and adaptive • What are these technologies, providing examples. How can AL EdTech, meant to provide learning technologies to adjust to the student? sufficient knowledge and help decisionmakers in the • What is the variability of the adaptability in these systems? understanding so that non- early stages on the process • How these technology can be implemented in a certain context? technical stakeholders can to explore the potential of make key planning, design, this technology as part of the Additionnal questions are addressed in the FAQ:* and procurement decisions solution for their pedagogical for education. problems in their context. DEFINING THE TECHNOLOGY • How are the World Bank’s 5 EdTech principals applied to They can be used as a starting personalized and adaptive learning? point for the planning of • Is there any evidence to support the problem statement? technology deployment to • What is the difference between adaptive learning, Computer improve education, especially Assisted Instruction (CAI) and self-directed learning? with education ministries. • What are the main system components of an adaptive learning software? • How do the algorithms work in an adaptive learning software? FOCUSING ON THE IMPLEMENTATION • Is there an initial checklist to analyze if personalized and adaptive programs can be part of the solution to a certain context? • Is there a checklist to select the most appropiate adaptive program for a certain context? • What are the next steps for a deployment process? *To know more about these questions and their answers, go to the FAQ available in the annex. 3 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES WHO are the main stakeholders ? RESPONSIBILITY KPs are designed with a human- • Assist MOE leadership in the application of KPs for in- Task Team Leaders (TTL’s) & country EdTech programs. Help design Bank-financed centered vision. Bank Project Managers projects with practical information to include in project (non-technical) documents. This knowledge pack is meant to provide sufficient MOE Leadership • Use KP to make key planning, design, and procurement knowledge and (non-technical) decisions for in-country EdTech programs. understanding to help decisionmakers make key planning, MOE Program Managers • Use KP to make key planning, design, and procurement design, and (semi-technical) decisions for in-country EdTech programs. procurement decisions of technologies for Donors, NGOs and Other • Use KP to align with Bank EdTech programs and establish personalized and Partners a common EdTech framework. adaptive learning. (non-technical) 4 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES HOW Problem statement | Possibilities | Evidence WHY is this KP designed ? PROBLEM STATEMENT EDUCATION CRISIS AND THE POTENTIAL OF ADAPTIVE LEARNING Learning gaps are cumulative Quality of teaching is the most and can lead to dropout. important factor contributing to student achievement. The number of students per teacher has constantly risen, Pedagogical interventions that which makes personalized promote personalized learning instruction difficult and/or and teaching at the right level expensive to implement. are effective.  5 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES HOW Problem statement | Possibilities | Evidence WHY is this KP designed ? POSSIBILITIES OF THIS TECHNOLOGY PERSONALIZED AND ADAPTIVE LEARNING AS PART OF THE SOLUTION Technology for personalized and adaptive learning, when integrated correctly, can be a solution to a difficult pedagogical problem: teaching at the right level of every student in a large and diverse classroom, at scale. Technology can deliver students personalized and adaptive content that adjust to their learning needs. The use of different technologies to foster personalized and adaptive learning has shown to be cost-effective and scalable. Promising outcomes Close educational gaps Personalization The use of technologies for Technology-supported personalized The adaptability of these technologies Personalized and Adaptive Learning learning may be most beneficial in allows ‘teaching at the right level’, appears to offer significant promise closing educational gaps for lower enabling students to learn at their to improve learning outcomes, attaining students, potentially own pace and according to their including potentially ‘out-of-class’ and including those returning to school current proficiency of the subject.  ‘out-of-school’ learning. after an absence. Source: adapted from the key findings from the “Technology-Supported Personalised Learning: A Rapid Evidence Review” report (June 2020), 6 EdTech Hub INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES HOW Problem statement | Possibilities | Evidence WHY is this KP designed ? EVIDENCE EVIDENCE OF WHAT WORKS AND WHAT DOESN’T According to the EdTech Hub “Technology-Supported Personalised Learning: A Rapid Evidence Review” report (June 2020), studies report diverse but broadly positive relationships between technology-supported personalized technology and learning outcomes: STUDIES POSITIVE OUTCOMES MIXED OUTCOMES NEGATIVE OUTCOMES RCTs 12 10 2 0 Quasi-experiments 8 5 0 4 Case study 4 3 0 1 Total 24 17 2 7 See more information about research and evidence in the following slides with examples (11,12,13). 7 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Structure of solutions | What has been done in other countries HOW WHAT are the potential solutions? STRUCTURE OF SOLUTIONS TYPOLOGY OF TECHNOLOGICALLY-ENABLED PERSONALIZED AND ADAPTIVE LEARNING SYSTEMS There is not one unique level of adaptation, but five: RESPONSIVE SYSTEMS ADAPTIVE SYSTEMS CUSTOM INTERFACE LEARNING MANAGEMENT DATA DRIVEN ADAPTIVE LEARNING ... INTELLIGENT TUTOR • Invites student to • Platforms that • Management • Data-driven learning • The image of a DEVELOPMENT CURENTLY IN personalize learning automate a range systems that that potentially proactive learning experience by of classroom provide materials moves beyond a guide, that could selecting colors and management tasks. appropriate to a pre-determined inspire questions, or avatars • Includes systems that students’ proficiency decision tree and use facial recognition allow students to level based on data uses machine to respond to choose their own path collection. learning to adapt to emotional states. through material. students’ behaviors and competency. Click for more info in the LMS KP Artificial Intelligence (AI) adapted Adjusted content according Choose content path to behaviors and competencies to performance level Range of personalization Source: Bulger typology, Personalized Learning: The Conversations We’re Not Having, https://datasociety.net/pubs/ecl/PersonalizedLearning_ primer_2016.pdf 8 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Structure of solutions | What has been done in other countries HOW WHAT are the potential solutions? STRUCTURE OF SOLUTIONS SOME EXAMPLES 1/3 BRIEF EXPLANATION RESEARCH & EVIDENCE OFFLINEABILITY LANGUAGE ED. LEVEL ALEKS is a research-based, online “spending approximately 30 h on a $30 learning program that offers course intelligent tutoring systems (ITS) was equally products for Math, Chemistry, Statistics, as effective as spending hundreds of hours ENGLISH and more.  and thousands of dollars of tuition”. Hickey, ALEKS Constantly adapting to update each Daniel T., et al. “Internet-based alternatives SPANISH student’s knowledge state, ALEKS for equitable preparation, access, and success K-12 students guides students to precisely what they in gateway courses.” The Internet and Higher Higher Ed are ready to learn at all times. Education 44 (2020): 100693. Includes text-based and video Alta makes an impact on learning outcomes for instruction, interactive learning content, students of all ability levels. Study of Knewton Alta assignments and review materials. Online Courses for Undergraduate Students: ENGLISH Knewton It can identify and dynamically Examining the Relationships Among Usage, boost knowledge gaps while you’re Assignment Completion, and Course Success. Higher Ed completing assignments. Wolf R. and all, 2018. Students receive hints and explanations “the intervention significantly increased student to assist their understanding as they scores on an end-of-the-year standardized complete their assignments. Teachers ASSIST- mathematics assessment as compared with ENGLISH get real-time assignment reports ENGLISH ments a control group that continued with existing detailing student and class homework practices.” (Roschelle, Feng, Murphy performance and common wrong & Mason, 2016). answers and other rich insights.  9 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Structure of solutions | What has been done in other countries HOW WHAT are the potential solutions? STRUCTURE OF SOLUTIONS SOME EXAMPLES 2/3 BRIEF EXPLANATION RESEARCH & EVIDENCE OFFLINEABILITY LANGUAGE ED. LEVEL An independent study funded by the U.S. Dep. of Students get the 1-to-1 feedback and Education, conducted by the RAND, the Carnegie encouragement they need and the Learning blended approach nearly doubled ENGLISH chance to own their learning and growth in performance on standardized tests Mathia monitor their progress, even as they relative to typical students in the 2nd year of SPANISH work from home. implementation. Pane JF et all, 2015. “Continued K-12 students Math. Progress. Promising Evidence on Personalized Learning”. This Digital Math Platform designed A study from the University of Western Sydney by education experts can work ENGLISH that shows Matific can help raise academic offline and is device agnostic, Matific performance by up to 34%. Attard, C. (2016). offering a wide range of possibilities SPANISH Research Evaluation of Matific Mathematics to be used in different environments Learning Resources: Project Report. K-6 students and contexts. Math. This is an adaptive learning program Studies carried out by independent evaluators that uses  AI and gamification Math- have shown a relationship between learning elements designed to support ENGLISH Whizz gains on Math-Whizz and increased performance students from kindergarten to grade in external assessments. Whizz Education Proof 0.5Mbps Internet 8 in math. connection (per Pack, July 2020. concurrent student) K-8 students 10 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Structure of solutions | What has been done in other countries HOW WHAT are the potential solutions? STRUCTURE OF SOLUTIONS SOME EXAMPLES 3/3 BRIEF EXPLANATION RESEARCH & EVIDENCE OFFLINEABILITY LANGUAGE ED. LEVEL Mindspark is a computer-based, online self- “[students using Mindspark] scored 0.37 standard Class 1-10 learning tool that helps children improve deviations higher in math and 0.23 standard Maths ENGLISH Minds- their skills in Math and English. It allows each deviations higher in Hindi over just a 4.5-month Class 4-9 park student to follow a learning path that is based period. Disrupting Education? Experimental Evidence HINDI English on their current level and at a pace they are on Technology-Aided Instruction in India. Karthik Class 6-8 comfortable with. Math, English, Science. Muralidharan, Abhijeet Singh, Alejandro J. Science Plataforma Adaptativa de Matemática is a digital online tool for students and teachers 0.20 standard deviations in Mathematics learning offered by Plan Ceibal in Uruguay. It provides gains in children who had used PAM compared to immediate feedback to the student after each students who had not used it. Higher effects were SPANISH PAM answer, offering help, theoretical materials observed in students of lower socio- economic and showing alternative solutions. Qualitative status. Perera, M; Aboal, D. (2018). The Impact of a studies suggest that communities of practice Mathematics Computer-Assisted Learning Platform on K-12 students and good connectivity at school influence Students’ Mathematics Test Scores. teachers’ decision to use PAM. Math. Read Along (called “Bolo” in the past) is an Google piloted the Read Along app in 200 villages in AI-powered reading app designed to help Unnao district, UP India from Oct ‘18 to Jan ‘19 with Hindi, English, Urdu, Bengali, primary grade kids learn to read. So far Bolo the help of operational support from the ASER Centre. Read Tamil, Telugu, has launched in Hindi, English, Urdu, Bengali, 64% of participants from the India pilot study with Marathi, Along Tamil, Telugu, Marathi, Spanish, Arabic and access to the app showed an improvement in reading After downloading Spanish, Arabic and Portuguese Portuguese. proficiency, and 92% of the parents noticed some the app, this can be K-6 students used offline. improvement in their child’s reading skills. In addition to the examples provided, the Edtech Hub, a research initiative supported by UK FCDO, the Gates Foundation and the World Bank, has 11 developed a list of different EdTech Tools that enable personalized learning INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Structure of solutions | What has been done in other countries HOW WHAT are the potential solutions? WHAT HAS BEEN DONE IN OTHER COUNTRIES - CASE STUDIES DOMINICAN REPUBLIC ECUADOR The World Bank team in region. Language was an Ecuador’s Secretariat of the students’ mathematical conjunction with the Ministry important challenge, and the Higher Education, Science, curricular improved between of Education of the Dominican cost of translating existing and Technology (SENESCYT), 8 and 10% per month, Republic (MINERD) designed and platforms and content can ver implemented an adaptative (equivalent to two full years implemented the Programate very high. computed assisted remediation of schooling), after using the project, which sought to • include resources for students, program in a 4 month platform for 16 consecutive improve learning outcomes in teachers and staff weeks. mathematics),carried out in • cost-effective: include pilot in 2021 (39 technical and • Costing: the cost to access the different stages: implementation technological institutes with platform from 10 to 20 USD capacity building for teachers more than 4,700 first-year per student per year (tutoring • A pre-pilot in 2018 (between students).  and remedial classes can March and May) 2-month pre-pilot in 2018 (420 The program was introduced in oscillate between 200 and 500 • Programate 2019 (from March 6th grade students in 5 schools), over 240 different classes with dollars per student per year).  to June) expanded to 51 schools. the support and participation of • Programate 2020 (from • Results: Students improved in 136 teachers: January to July) different areas • Objective: to contain the • Conclusion: positive increase in student dropout, Criteria to select the right relationship between the time partly explained by low adaptive software: spend in the software and the levels of student academic • available in Spanish results in math. readiness in mathematics. (experience in Latin American • Results: the knowledge of Adaptive Technology to Help Improve Math Learning in the Dominican Use of adaptive computer assisted remediation programs to prevent Republic student dropout in the context of COVID-19 12 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Deployment process | Costs and budgeting | Monitoring and Evaluation HOW to implement next steps ? DEPLOYMENT PROCESS - AREAS TO CONSIDER TO START IMPLEMENTING ADAPTIVE LEARNING PROGRAMS The use of technology for personalized and adaptive learning requires a series of enablers to use and adapt this alternative pedagogy in the classroom. Infrastructure Teachers Content and platform Policy • What is the minimum • Is there sustained guidance and • In what grades is going to be • Does funding exist? infrastructure required to training for educators on how to implemented? • Is there organizational experience implement certain adaptive program use the program? • Is the content aligned to the and expertise to drive the (electricity + connectivity + • Are the principals, teachers and curriculum? implementation? devices)? students on board? • Is it available in local/indigenous • Can the program be scaled up? • Has the hardware the right software? • Plan for a gradual incorporation languages? • Is the team planning to measure • Does the adaptive program provide of the tool: using these type of • Is the content culturally relevant? the impact of technology on technical support? programs changes the dynamics of • Does the vendor allow to utilize learning and other variables? the class. locally produced resources? • Are data privacy protection policies being considered? It is important to consider the most Adaptive systems are not meant to Adaptive systems require a detailed District education leaders need to be appropriate software according to the replace teachers, but rather enhance curriculum mapping and content able to monitor/track the progress of context of the country, including those their role. Communities of practice can development. their designated schools and support that are device agnostic (that work in play an important role. as well as hold accountable the school different devices) and those that can Policymakers can adapt a pre-existing, leaders to implement such initiatives in work offline. Teacher training, development and proprietary system for their context their schools. capacity building (digital, pedagogical, (developing a system from scratch is Equity. PAL solutions have a lot technical skills) should be an ongoing the least preferable option). Good use of the collected data must of potential, but they are heavily planned development program. be ensured, collaborating with other dependent on having access to digital agencies, as indicated in this UNICEF’s device, potentially exacerbating the Another interesting area to explore report. digital divide. is capacity building for parents/ caregivers to support children using 13 these tools. INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Deployment process | Costs and budgeting | Monitoring and Evaluation HOW to implement next steps ? COSTS AND BUDGETING Depending on the context, the biggest cost on these interventions can be upfront costs, such as hardware and connectivity infrastructure. QUESTIONS COST ELEMENTS REDUCING THE COST ELEMENTS • Importance of marginal cost, the cost of allowing • The expertise/implementation layer is relatively an additional student to use the technology. This fixed and at scale is a small percentage of overall • Does funding will be different in different countries. exist? cost, yet highly impactful. • Is there existing • Content development, curation and production • Curate existing content instead of developing new infrastructure (including research, translation and adaptation content. that could be to local languages and contexts): • Work in coordination with the platform provider used to reduce to • Vendor resources to allow using already created content, instead of cost of starting/ • Adaptation of other resources (including expanding this creating all from scratch. program? OER) • Try to get a better price per license based on scale • Is it going to be • Copyright fees including maintenance and updates. implemented in • Student Licenses • Some adaptive learning systems are increasingly schools? • In-the field support capacity device agnostic so can work on smartphones, • Is there • Technical (responsiveness to connectivity organizational which aligns with “bring your own device” policies. experience and hardware issues) • The upfront cost of investing in building local field and expertise • Faculty training (to build familiarity and support capacity for schools and communities sets to drive the confidence with the platform) up for lower overall long-term cost. implementation? • Educational/pedagogical (to allow integration • In those countries were there are some computers • Can the program with schools’ learning and teaching) be scaled up? in some schools, if those can be used, the cost of expanding/starting this program is lower. 14 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Deployment process | Costs and budgeting | Monitoring and Evaluation HOW to implement next steps ? MONITORING AND EVALUATION It is essential to monitor the implementation and measure the impact of technology on learning, in order to understand if the intervention is achieving the proposed objectives and to correct based on the lessons learned (course correction). Based on the 4 areas previously introduced in this KP, it will be important to ask some questions to better understand the impact, e.g.: Infrastructure Teachers Content and platform Policy • Was the technology • Were teachers trained on • Was the content mapped • Are national/regional adequate to the currently how to use the technology, to the local curriculum? education leaders able available infrastructure and in how to incoporate • Was the content to monitor/track the in the context (in terms of the tool effectively in the contextualized? progress? electricity, connectivity classroom? • Was MoE staff trained to and devices)? work with such systems • Were teachers trained on and were policies adapted • Equity: How did the how to utilize the data accordingly, providing the program dependency on and findings to Support Ministry with the ability to having access to digital personalized learning? make evidence-informed devices. Impact the policy decisions (data- existing digital divide of the driven)? students? 15 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES HOW Conclusion WHAT • The use of technology for personalized and • Technology can deliver students adaptive learning requires a series of enablers, personalized and adaptive content such as infrastructure and teacher readiness, to that adjust to their learning needs. use and adapt this alternative pedagogy in the classroom. The high dependency on devices could • Adaptive systems are not meant to potentially exacerbate the digital divide. replace teachers, but rather empower them, providing them with tools to • Algorithms can only measure what they are teach at the right level and supporting programmed to measure. It is important to them to be more responsive to the understand the limitations of the different technologies and have clarity in the outcomes. specific needs of each student in the Good use of the collected data must be ensured. classroom. PROS • Further research is needed to investigate the CONS • The use of different technologies for various complex and nuanced factors associated personalized and adaptive learning with technology-supported adaptive learning. has shown to be cost-effective and scalable. • It requires time (It require also time, institutional support, technical assistance, incentives, etc.) 16 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES HOW Conclusion WHAT WHO WHY In order adaptive and personalized learning Personalized and adaptive learning programs, programs to work, principals and teachers must when integrated correctly, can be a solution to a be on board and trained, and students need to difficult pedagogical problem: teaching to the be engaged. Adaptive systems should empower right level of every student in a large and diverse teachers, rather than replace them. classroom. The use of different technologies for personalized and adaptive learning has shown to be cost-effective and scalable. Adaptive systems can provide personalization The deployment process should include three phases: at scale, using a data-driven  approach to Design, Deploy, and Sustain. It is essential to WHAT instruction and remediation. This technology understand the problem, identify if the use of adaptive HOW can dynamically adjust to student interactions and learning technology could be part of the solution and performance levels, delivering the types of content study the best alternatives based on the context. in an appropriate sequence that individual learners Before the implementation, consider the status of need at specific moment to make progress. the enabling conditions (at the infrastructure, content, capacity and policy levels). 17 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES To go further CLOUD OF KPs R E L AT E D S O U R C E S AI/ML Devices Cloud Adaptive Blog: Considering an adaptive Learning Connectivity learning system? A roadmap for Digital Infrastructure policymakers Digital identity BE DATA Data visualization DRIVEN ENGAGE the NRENs Podcast: Adaptative Learning ECOSYSTEM Procurement Podcast Ecosystem Startups Data collection EMIS AssistiveTechnologies LMS LEARNER DESIGN and Mobile based S T AY C O N N E C T E D Computer based ACT AT SCALE, Digital ASK WHY? FOR ALL Assessment Literacies Follow us on Twitter Digital Content EMPOWERED TEACHERS Subscribe to our podcast channel Teachers Competencies Spotify & Anchor More updates on Medium Subscribe to our EduTech Newsletter OTHER EXISTING RELATED KPs Assessment LMS Procurement EdTech website Digital content Teachers’ competencies 18 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT GLOSSARY AI (Artificial Intelligence):  A discipline neurological processes of teaching and ML (Machine Learning): A branch of artificial learning, offering something much closer to concerned with the building of computer intelligence concerned with the construction an interactive textbook than a tutor. (Source: programs that perform tasks requiring of programs that learn from experience. Personalized Learning: The Conversations intelligence when done by humans. (Source: (supervised, unsupervised, reinforcement We’re Not Having). A Dictionary of Computer Science). There learning). (Source: Adaptation from A are different AI technics (including machine Dictionary of Computer Science). learning and neural networks –including deep Algorithm: A prescribed set of well-defined learning-, among others).  rules or instructions for the solution of a problem, such as the performance of a Adaptive systems / Responsive systems: calculation, in a finite number of steps. (Source: A Dictionary of Computer Science).  Adaptive systems aim to functionally mirror and support the learning process, which is a flexible and changing, rather than fixed, LMS (Learning Management System): An LMS process. Responsive systems are more limited, essentially offering an interface is at the core of an eLearning system and has to predetermined content, like a hyper- to interact with several other systems that can linked menu or a series of digital buttons. be integrated or standalone. In comparison to truly adaptive systems, responsive systems are further from the Click for more info in the LMS KP 19 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT REFERENCES AND LINKS OF INTEREST • Considering an adaptive learning system? A roadmap for • Personalized Learning: The Conversations We’re Not Having https:// policymakers https://blogs.worldbank.org/education/considering- datasociety.net/pubs/ecl/PersonalizedLearning_primer_2016.pdf adaptive-learning-system-roadmap-policymakers • Technology-Supported Personalised Learning: A Rapid Evidence • Use of adaptive computer assisted remediation programs to BLOGS Review, by the EdTech Hub, https://docs.edtechhub.org/ prevent student dropout in the context of COVID-19 https:// lib/?all=major+pal&page=1&page-len=1&sort=score&id=A2II5ZV7 blogs.worldbank.org/education/use-adaptive-computer-assisted- • The effectiveness of technology-supported personalised learning remediation-programs-prevent-student-dropout-context-covid in low- and middle-income countries: A meta-analysis, https:// • Adaptive Technology to Help Improve Math Learning in the docs.edtechhub.org/lib/?all=major+francis&page=1&page- REPORTS Dominican Republic https://blogs.worldbank.org/education/ len=1&sort=score&id=5U948655 adaptive-technology-help-improve-math-learning-dominican- • Artificial Intelligence in Education: Promises and Implications republic for Teaching and learning. https://circls.org/primers/artificial- intelligence-in-education-promises-and-implications-for-teaching- and-learning • The Case for Better Governance of Children’s Data: A Manifesto. PODCASTS https://www.unicef.org/globalinsight/reports/better-governance- childrens-data-manifesto • Mitigating Learning Losses and Accelerating Learning • Technology-mediated personalised learning for younger learners: through Adaptive Learning https://open.spotify.com/ Concepts, design, methods and practice https://bera-journals. episode/6IqG04BaEuzzmbAouKiMAG onlinelibrary.wiley.com/doi/10.1111/bjet.13150 20 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ DEFINING THE TECHNOLOGY FOCUSING ON THE IMPLEMENTATION How are the World Bank’s 5 EdTech principals applied to Is there an initial checklist to analyze if personalized and adaptive personalized and adaptive learning? programs can be part of the solution to a certain context? Is there any evidence to support the problem statement? Is there a checklist to select the most appropiate adaptive program for a certain context? What is the difference between adaptive learning, Computer Assisted Instruction (CAI) and self-directed learning? What are the next steps for a deployment process? What are the main system components of an adaptive learning software? How do the algorithms work in an adaptive learning software? 21 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ Back to DEFINING THE TECHNOLOGY How are the World Bank’s 5 EdTech principals applied to personalized and adaptive learning? ASK WHY?: ENGAGE THE ECOSYSTEM: If technology is the answer, what is the Education systems should take a whole- rules related to data privacy, ownership, usage question? Considerations of the use of adaptive of-government and multi-stakeholder approach, and security. Program stakeholders can also use learning technology should focus on “education”, both inside and outside the system. They the date for continuous improvement, targeted on the pedagogical problem to solve and not must bring together stakeholders, like various investment, and to inform policy. just on the “technology”. It is essential to set very ministry departments (such as the curriculum clear objectives with clear learning outcomes department, the teacher development measurements department and the examination board), educators, adaptive software companies, DESIGN AND ACT AT SCALE, FOR ALL: implementation capacity/partners with expertise EdTech interventions must be designed for and contextual experience, content creators, scale for all children. infrastructure partners, and local EdTech Adaptive systems can provide personalized startups. learning at scale. Click to read the EdTech DATA DRIVEN: EMPOWERED TEACHERS: Valuable information can be collected Strategy Technology should enhance teachers’ in adaptive learning programs regarding the access to content, data and expertise to improve differences in the learning experience. This data teaching and learning. To be deployed effectively, presents an opportunity for teachers to better adaptive systems must include training for cater to student’s learning gaps and to inform educators. decisions to improve teaching and learning. This must be combined with clear policy guidance and 22 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ Back to DEFINING THE TECHNOLOGY Is there any evidence to support the problem statement? • The world is in the midst of a technological »» Individualized, repeated teacher training, revolution. Students are not being adequately associated with a specific method or task prepared to thrive in this rapidly changing world. • Teaching in developing country contexts is • Education systems, especially in low- and- difficult: Hight student-teacher ratios, Teaching middle-income countries, face many daunting to the top of the distribution (Banerjee and challenges. 258 million students are out of Duflo, 2011; Glewwe et al. 2009; Glewwe & Personalized and school, including 59 million children of primary- Muralidharan, 2015; Pritchett and Beatty 2012), adaptive learning school age. The situation is even worse in Curricula in developing countries originally programs, when communities afflicted by conflict and violence. designed to screen gifted students for positions of integrated correctly, Girls and children with special educational needs responsibility, etc. are particularly being left behind. This learning can be a solution to a crisis has been exacerbated by the health and • The number of students per teacher has difficult pedagogical economic crisis of COVID-19. constantly risen for the last decades (Nicol problem: teaching & Macfarlane‐Dick, 2006), making impossible to the right level of • The most important factor contributing to student to provide an individualize support to the achievement: quality of instruction (Hiebert and students. This leads to poor learning outcomes, every student in a Grouws, 2007; Rowan et al., 2002). Most effective high drop out rates and dissatisfaction (Brinton large and diverse interventions at increasing student learning et al., 2015; Eom, Wen, & Ashill, 2006; Hone & El classroom, at scale. are concerned with improving the quality of Said, 2016). instruction (Evans and Popova, 2016): • Gaps in the knowledge of curricula are »» Pedagogical interventions that match cumulative and can lead to the dropouts. teaching to individual student learning levels 23 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ Back to DEFINING THE TECHNOLOGY What is the difference between adaptive learning, Computer Assisted Instruction (CAI) and self-directed learning? DEFINITION of ADAPTIVE LEARNING COMPUTER ASSISTED INSTRUCTION / LEARNING It makes possible to adapt and/or redesign learning Computer Assisted Instruction (CAI), also sometimes materials for each individual learner. Taking different referred to as computer-assisted learning (CAL), uses parameters (such as student background, performance, computers together with traditional teaching. Computer- goals, abilities, skills, and characteristics) into consideration, assisted training methods use a combination of multimedia adaptive learning tools allow education to become more such as text, graphics, sound, and video in order to enhance personalized and student-centered than ever before. The level of learning. The primary value of CAI is interactivity – it allows “adaptation” varies from doing an initial assessment to determine students to become active learners instead of passive learners. a fix set of content and a fix learning path, to use large data pools and analytics to update the learning path by adapting the SELF-LED/SELF-DIRECTED LEARNING content and instructional strategy in real time to maintaining a ’Self-directed learning’ describes a process by which individuals complete learning profile for each learner. These systems offer take the initiative, with or without the assistance of others, in different levels of personalization based on intensive use of data: diagnosing their learning needs, formulating learning goals, the more data, the more personalized the experience will be. identifying human and material resources for learning, choosing and implement appropriate learning strategies, and evaluating learning outcomes. There are self-led learning programs that are not adaptive and vice versa. 24 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ Back to DEFINING THE TECHNOLOGY What are the main system components of an adaptive learning software? Some of the key components include: the curriculum (modules of the content to be learned by the students), the resources and activities to increase student’s engagement and interaction and the assessment (that creates datapoints about the student’s progress and provides feedback, linked to the next activities presented to the learner) and a database with the students’ information. Source: prepared by author based on a presentation by the Arizona State University (ASU) 25 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ Back to DEFINING THE TECHNOLOGY How do the algorithms work in an adaptive learning software? Looking into the “black box” of adaptive learning systems, including: • Regular flow feedback to the system and users • Personalized sequencing of the content • Individualized pace of learning & regulation of the cognitive load Source: adaptation by author based on a Dreambox learning scheme 26 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ Back to FOCUSING ON THE IMPLEMENTATION Is there an initial checklist to analyze if personalized and adaptive programs can be part of the solution to a certain context? Click for other KPs Depending on the context (existing infrastructure, available content, capacity Click for other building…), digital infra. KPs adaptive technology solutions may Click for other content KPs work on the country or not. 27 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ Back to FOCUSING ON THE IMPLEMENTATION Is there a checklist to select the most appropiate adaptive program for a certain context? • Provides adaptive and individualized learning • Includes diagnostic evaluation Adaptiveness • Offers diagnosis and mapping of knowledge per student and per course • It allows to identify and work on the topics that the student is ready to learn • Provides detailed progress & performance metrics for the knowledge acquired • The content is available in the local languages, and it’s contextualized Content • Allows the customization of the topics of a course with content from different courses/books • Includes textbook with links from the exercises Check the boxes • Allows teachers to create their own exercises Teachers • Includes a teacher module • It is compatible with various mobile devices Infrastructure • Works with internet speeds less than 1 mbps • Works offline • Offers motivating messages when the student learns a new topic Engagement • Provides other gamification elements 28 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Glossary | References | FAQ HOW Annexes WHAT FAQ Back to FOCUSING ON THE IMPLEMENTATION What are the next steps for a deployment process? DESIGN DEPLOY SUSTAIN • Establish Project Management Team • Request For Information (market • Capacity Building and Skills • Vision, Goals and Objectives consultation) Development • Define Use Cases • Define Technical specifications • Governance & Compliance • Evaluate Readiness and • Define Software stack • Partner & Stakeholder Management interdependencies • Request for Proposal • User Support • Identify appropriate devices • Procurement • Maintenance • Adoption & Change management • Delivery • Decommission • Budget (considering TCO) • Distribution • Replace / update • Risk & Mitigation • Set-up • e-waste management • Communication CONTINUOUS MONITORING AND EVALUATION OF PROJECT KPI’S 29 INTRODUCTION WHO WHY WHAT HOW CONCLUSION ANNEXES Supported with funding from