AI-Powered Learning Environments Framework PDF

Framework for AI-Powered Learning Environments This framework is intended to provide education leaders a strategic approach for integrating artificial intelligence into their systems based on best practices from successful implementation in other sectors. This work may be cited as: The National Ce...

Framework for AI-Powered Learning Environments This framework is intended to provide education leaders a strategic approach for integrating artificial intelligence into their systems based on best practices from successful implementation in other sectors. This work may be cited as: The National Center on Education and the Economy, Framework for AI-Powered Environments, (Washington, DC: National Center on Education and the Economy, 2024). Copyright © 2024 by The National Center on Education and the Economy ®. All rights reserved.  The quick and widespread presence of generative artificial intelligence (GenAI) and the ease of access afforded by ChatGPT has disrupted the education landscape and brought about a new wave of unprecedented challenges for educators. Some have responded with resistance to preserve continuity for students as they make sense of this new tool, while others have jumped in fully. Regardless of the approach, it’s time to develop a strategic view of how to integrate GenAI into our classrooms, schools, and entire education system. How should The question is no longer “Should we give students access to AI?”, but rather, “How should students work with AI?”— as most learners students work already have access to it, and the pace will only accelerate. The time is here for education leaders to affirmatively drive these changes and with AI? ensure their safe implementation in schools. This framework is intended to provide education leaders a strategic approach for integrating artificial intelligence capabilities into current education systems in the near term (18–20 months) and for the longer term (two to six years). It is based on best practices from successful implementation in other sectors. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 1  It’s ChatGPT! It has been more than a year since the Finally, it might be argued that teachers and public release of ChatGPT in late November students are leading the way in the use of 2022 took much of the world by storm. In GenAI. Teachers are flooding social media with the months following the initial release, a tips, sharing prompts for lesson development trickle of documents gave way to a deluge of and other resources, and collaborating with publications, webinars, websites, and tools peers through newly energized networks. providing guidance on use of GenAI within the Meanwhile, more than half of students in education sector. grades 9 through 12 claim to have used it for both personal and school purposes.3 And For example, the OECD’s report, the incorporation of AI into popular social Opportunities, guidelines and guardrails for platforms used by Gen Z and Gen Alpha effective and equitable use of AI in education1, suggest that AI-driven informal learning presented a set of key recommendations outside of the classroom is happening faster and questions for further exploration about than ever. how GenAI can support learning, teaching, and assessment. Earlier in the year, Artificial Intelligence and the Future of Teaching and Learning2 and other notable publications, including TeachAI’s AI Guidance for Schools Toolkit and ISTE’s Artificial Intelligence in Education examined similar issues. Large education technology companies entered the space with creative initiatives that use GenAI for individual student support, with Khan Academy’s Khanmigo and HMH’s Amira Learning being two early examples. And there has been a rising tide of entrepreneurial start- up companies finding creative ways to apply GenAI and other machine learning processes to help address educational opportunities and challenges. Notably, within most of the high- performing jurisdictions that NCEE studies, there has been a concerted effort to integrate generative artificial intelligence and machine learning across the entire system, especially in Singapore, Estonia, and Korea. 1 OECD, 2023. 2 2 U.S. Department of Education, Office of Educational Technology, May 2023. 3 How Students Use AI vs. How Teachers Think They Use It, in Charts, Ed Week, September 29, 2023.  It’s more than ChatGPT! The current landscape in AI is transitioning from GenAI 1.0 to GenAI 2.0. Initially, GenAI 1.0 centered around leveraging foundational models like GPT-4 (OpenAI), Llama 2 (Meta) or PaLM2 (Google AI) to develop applications and use cases built on these platforms. However, this approach is rapidly evolving. In GenAI 2.0, there is a significant shift, particularly in sectors outside of education such as the global steel industry, financial services, pharmaceuticals, and country-level energy production. The trend is moving to a more logical mix of technologies —a move from light touch use to more embedded applications. In other words, GenAI models aren’t used as the sole technology but are embedded in end-to-end workflows. Attempting to forecast the future of AI involves a high degree of uncertainty and complexity. Nonetheless, it’s insightful to consider the current market signals and their implications. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 3  Emerging Cross-Sector Trends in AI Predicting the future of AI is challenging and complex due to many unknowns. However, it’s valuable to look at current market signals and what they might mean for the future. Overall, these emerging trends are sure to have implications for the education sector, influencing the why, the what, and the how of teaching and learning. Moving from a sole reliance on foundation models such as Large Language Models (LLMs) to the use of GenAI in discipline- or organization-specific, customized models, leveraging their unique data sets and documents. This not only enhances data security by minimizing exposure, but also optimizes the use of GenAI for more tailored applications. This approach allows for discipline specificity; reduction in hallucinations (i.e., the fabrication of erroneous information); increased compliance with existing (e.g., the EU’s general data protection regulation, or GDPR) and emerging policies on data privacy; and energy efficiency, with an eye toward sustainability. Integrating advanced AI models into regular workflows, making them an essential component of daily operations and decision-making processes. This strategic technological mix often combines predictive analytics for process improvement, digital twins for modeling and testing, and GenAI to elevate design and productivity. Reshaping organizational structures and cultures—extending beyond the dynamics of hybrid (remote/on-site) models. Embedding GenAI tools across all levels from management to frontline operations can improve decision-making capabilities, both individually and collectively. This has the potential to promote greater autonomy throughout the organization, boost team effectiveness, and increase productivity. Additionally, it may foster a more human-centered workplace. All of these have serious implications for leadership in a modernized, next gen world of work, as well as many of the post-secondary environments that students are likely to enter. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 4  Underlying Assumptions Given the wide spectrum of hopes and fears accompanying the role of AI in our society, it’s important to clearly identify our underlying assumptions regarding AI’s impact on both education and society as a whole. At the core of this framework are the following assumptions: Despite the inherent risks, GenAI has the potential to significantly benefit humanity. Integrated with other existing and emerging technologies, AI can help improve lives in the way that many similar technological leaps have in the past. For example, there are ongoing cross- sectoral initiatives exploring AI’s capacity to foster a more sustainable future, particularly in areas such as energy generation, water access, and food production—as well as driving more equitable employment opportunities across diverse workforce segments and improving overall living standards. Artificial Intelligence is poised to substantially transform educational systems and learning environments. In the near future, PK–12, higher education, and post-secondary work environments will look very different from their current forms. This also extends to a wide array of applications for self-directed, continual learning. The emergence of GenAI offers the possibility of closing gaps in outcomes for students — particularly those who have not succeeded with a “one size fits all” model, or from communities who lacked access to new technologies and private educational resources. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 5  Guiding Principles AI represents a significant educational and societal disruptor, but it is not the first, nor will it be the last. It is important to remain cognizant of fundamental guiding principles when addressing any form of disruption to our systems. While the scale of AI’s impact may be unparalleled, the concept of disruption is familiar. The principles outlined below are applicable to a range of innovations and disruptions, whether technological or otherwise, ensuring a structured and effective response. A “North Star” vision. Any new innovation or disruption must fit into a community’s vision for education. A shared understanding of the goals, purposes, and value of education is essential, acting as a “North Star.” Equitable access. Guaranteeing that all students, without exception, have equal access to these advancements is a fundamental ethical obligation of public education. In this context, the term “students” explicitly includes every student. It is especially crucial to provide these opportunities to those who are furthest from opportunity and have not benefited from technological advancements. Human-centeredness. Research around the science of learning continually underscores that we learn through social interaction. With any new innovation or disruption the roles of teachers and the leaders who support them become more, not less, important. In this era where AI-driven systems provide significant scaffolds to learning, teachers will still be relied on to help students develop the capacity to ask meaningful questions, think critically, engage in discussion for deep understanding, and guide collaborative learning among groups of students. Conditions of safety (privacy, data security, and the appropriateness of content) and a culture of ethical decision-making (responsible usage). These factors, while particularly prominent with GenAI, are crucial when assessing, acquiring, or applying any new innovation or disruptive change. A strong focus on sustainability. This includes considerations such as carbon footprint, water usage, and the broader environmental impact of manufacturing hardware and infrastructure. Companies involved in AI development must strive to innovate while also aiming to meet ambitious net-zero targets. Those using it—including districts and schools—have a responsibility to ensure these companies are accountable for their environmental commitments. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 6  About the Framework The framework is organized across two time horizons: Now—end of 2025: focuses on the application of emerging AI tools to current education systems in what might be thought of as a “learning phase.” The emphasis is on aligning the use of AI models with the science of learning, as well as emerging views of post-secondary pathways and experiences and the ethical use of AI. 2026—2030: focuses more on the integration of technologies into the entire education system along with ways to scale and optimize technologies to transform the system itself. It also recognizes that rapid technological advancement is changing what students need to know and be able to do and asks how the education system needs to change to prepare students for this future. The prioritization within each time horizon is flexible depending on the context. The framework is designed to be adaptable, catering to the needs of any leader or district, regardless of their current state. Some leaders might find themselves aligning with the second horizon, while others are still preparing for the first. The goal is to provide support irrespective of where one stands in the spectrum. And given the rapid pace of technological change, this framework will be updated periodically to ensure its relevance and usefulness to education leaders. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 7  AI Framework Near Term: Now—End of 2025 Horizon 2 —2030 Longer Term: 2026­ Districts and schools need to raise awareness Education systems must set high expectations and create visibility into the power and for students, focusing on the knowledge, usefulness of GenAI by investing in and skills, and dispositions necessary to succeed encouraging baseline training for students, in a world marked by constant change and teachers, district leaders, parents, and other technological advancement, and develop key stakeholders. performance assessments that measure them. Every student and teacher should be able Education systems must be significantly to use AI tools for knowledge access and modified to align to newly-defined high management. Schools are encouraged to expectations, resulting in a dramatic change integrate these tools into all types of learning to how students acquire knowledge and skills, experiences including core academics, co- how they work together, how they apply their curricular, and career and technical programs. learning, and how they develop agency. Student experiences should be structured so that learning is authentic, relevant, and Teachers remain essential. Their role must meaningful, using the latest and most advanced evolve from helping students acquire technology. knowledge and skills to becoming facilitators of inquiry and guiding students as they apply their Students need to develop a strong command learning to complex tasks. of using language (both English and their first language if it’s different) including writing, Leaders should create a culture of continual reading, speaking, and thinking logically innovation and renewal, while ensuring a as defined by college- and career-ready balance between introducing new ideas and standards. These skills enable them to most maintaining stability within the system. effectively work with AI-powered machines. Aligned instructional systems should utilize the power of AI to provide adaptive instruction that scaffolds a student to reach grade-level proficiency (or beyond) based upon a student’s interactions with the system. Teachers must be supported by organizational structures and incentives that reflect the modern reshaping of professions that drive the knowledge and creative economies. GenAI models need to be used to increasingly support leadership responsibilities. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 8 Near Term: Now—End of 2025 Districts and schools need to raise awareness and create visibility into the power and usefulness of GenAI by investing in and encouraging baseline training for students, teachers, district leaders, parents, and other key stakeholders. Learning about AI—getting to know how AI works and how to use it—is becoming more and more important for students, teachers, parents, and all stakeholders. A report from the World Economic Forum (December 2023) gave businesses advice to help their employees understand and learn about AI and large language models (LLMs). It encouraged business leaders to...introduce courses that explain LLMs, including their limitations and how they can be used, to demystify the technology for workers. By establishing a baseline level of LLM fluency among their entire workforce, companies can help their employees understand what LLMs are, how their jobs will evolve to work alongside the technology, and its benefit to the enterprise when responsibly deployed. This approach not only builds confidence in LLMs but also encourages adoption and use.4 This approach is equally important within the education sector. Many of the best education systems that NCEE studies have started programs like this. It matches a key guideline promoted by the OECD and Education International about using AI in education. They suggest that jurisdictions work with end-users to create AI learning tools. They state that [j]urisdictions should encourage the involvement of teachers, students and other end users as co-designers in the research and development process of technology to help ensure the usefulness and use of AI-enabled digital tools.5 Supporting a foundation of AI-literacy for all stakeholders is the surest way to support this approach. Where It’s Happening: Estonia added learning modules to a platform, ProgeTiger, which has introduced students from early ages to programming and robotics for the past ten years. New modules added in 2023 focus on topics such as how AI is used by professionals in different sectors (including education) and the ethical dilemmas of AI—all while helping to guide students to successfully use AI tools. United States, Florida Department of Education has launched a career and technical education program, Artificial Intelligence Foundations, a sequence of courses developed in partnership with the University of Florida, based on their Artificial Intelligence for K-12 initiative. The program is designed to promote AI literacy and to help students to become AI-ready for their post-secondary experiences. Singapore adopted a national strategy in artificial intelligence, NAIS 2.0: AI for the Public Good for Singapore and the World. Underpinning the strategy is the vision that Singapore will be a place where AI can uplift and empower people and businesses to “thrive in an AI-enabled future.” One of the goals in support of that strategy is to equip individuals and communities “to use AI with confidence, discernment, and trust.” To support this goal, Singapore has invested in a learning platform LearnAI Singapore which supports students, educators, and professionals working in fields outside of education with AI literacy. 4 Jobs of Tomorrow: Large Language Models and Jobs – A Business Toolkit, WEF, December 2023. 9 5 Towards an Effective Digital Education Ecosystem, OECD Digital Education Outlook, 2023. Near Term: Now—end of 2025 Every student and teacher should be able to use AI tools for knowledge access and management. Schools are encouraged to integrate these tools into all types of learning experiences including core academics, co-curricular, and career and technical programs. Student experiences should be structured so that learning is authentic, relevant, and meaningful, using the latest and most advanced technology. There is strong emphasis on the importance of all students having access to the latest and best digital learning tools, including those with advanced AI features.6 This need grows with the evolution of new tools that improve how we teach and learn. And in the workforce, artificial intelligence—especially GenAI 2.0—is boosting productivity in the knowledge economy and creativity space. This boost comes from AI being trained on vast data sets.7 Large Language Models (LLMs), for instance, are becoming more focused, explainable, and traceable through integrated applications that allow the use of local data and documents (user-input) without mixing it with public data. This advancement allows us to leverage the extensive information available in the digital universe and spot “weak signals” in specific knowledge areas. Learning should be real, interesting, and useful for students to help them build important skills like critical thinking, creativity, and learning throughout their lives. This isn’t just about using AI or other technology. In order to help students develop the skills necessary to take command of their respective futures, not only must students have access to the real world tools used in modern work, they need to be able to apply these tools within the context of authentic tasks. This also means that teachers need to understand how to use these tools and integrate them into the learning environment. Integrating AI into student learning environments indicates a needed shift in performance tasks and assessments to be more trans-disciplinary, collaborative, and applied. This approach will accomplish two goals: first, it makes sure that AI can’t do the work for students, so they have to put things together and create something themselves. Second, it prepares students for the kind of work they’ll likely do after school, in jobs or further studies. Within this context, there are two important considerations: Students, teachers, and administrators must understand the ethical issues and risks associated with AI systems and outputs, especially with regard to GenAI models.8 Students should be encouraged to use AI tools just like experts do, and they should be given explicit guidance on how to use them effectively. For example, students can use these tools to come up with ideas for projects, write the first drafts of their work, get feedback, check if their way of solving problems works, or make summaries. 6 See Guideline 2, p. 7 in Opportunities, Guidelines and Guardrails for Effective and Equitable Use of AI in Education, OECD, 2023. 7 Initial dataset for GPT-4 was about 1 petabyte of data comprising web texts, books, news articles, social media posts, code snippets, 10 and more. GPT-4 Everything you want to know about OpenAI’s new model, 3/14/23; accessed 4/4/24. 8 Ensuring a Responsible Approach to AI and Companion Guide: Ensuring a Responsible Approach to AI from Code.org’s toolkit, AI 101 for Teachers lays out a set of discussions around data privacy, hallucination, and bias. ISTE’s Bringing AI to School: Tips for School Leaders summarizes many of these issues while providing links to other resources. Near Term: Now—end of 2025 Students need to develop a strong command of using language (both English and their first language if it’s different) including writing, reading, speaking, and thinking logically as defined by college- and career-ready standards. These skills enable them to most effectively work with AI-powered machines. The foreseeable future points to only more collaboration between humans and machines, increasingly driven by natural language. In fact, human-computer interactions are taking advantage of recent advances in speech-to-text and text-to-speech, so the ability to communicate through both written and oral language is critical. In the near- and mid-term, students will communicate with AI models via written text or speech (and the machine will increasingly be able to accommodate dialects, accents, and inflections); the AI model will “talk back” in both formats (written text or speech); iterative interactions with AI models can help guide the use of precision in language and hone students’ logical thinking. These interactions rely on a precise use of natural language. The ability to query and provide direction to AI models with a clear understanding of logic, problem definition, and atomization, as well as the ability to test claims, evidence, and warrants will be critical to their productive use. Thus, strong proficiency with language—writing, reading, and speaking—will become even more of a gateway to many jobs of the near future. Teachers’ continual reinforcement of this process (along with their understanding of how natural language is likely to evolve in this capacity) will support students with the development of this core competence. Where It’s Happening: The curriculum and instruction for supporting this already exist within current systems. In the United States, a majority of state standards are built from or align to the Common Core State Standards, which themselves are internationally benchmarked. As such, they have the following (or similar) language already in place: Students who are college- and career-ready in literacy “demonstrate independence. Students can, without significant scaffolding, comprehend and evaluate complex texts across a range of types and disciplines, and they can construct effective arguments and convey intricate or multifaceted information. Likewise, students are able independently to discern a speaker’s key points, request clarification, and ask relevant questions. They build on others’ ideas, articulate their own ideas, and confirm they have been understood.9 Those that have more recently revised their standards will have similar statements, to which the standards for English Language Arts will be aligned. 9 Common Core State Standards ELA-Literacy 11 Near Term: Now—end of 2025 Aligned instructional systems should utilize the power of AI to provide adaptive instruction that scaffolds a student to reach grade-level proficiency (or beyond) based upon a student’s interactions with the system. Adaptive instructional models are like personalized teaching tools. They deliver targeted core content, modified tasks, ancillary resources, and “in the moment” coaching and feedback to a student based upon what the system has learned about the student’s level of proficiency by their use of the system. This ability to adjust to each student’s needs is a powerful benefit of AI in education. It’s not just about personalization and student agency over their learning—it’s also about understanding in real time where a student is within a proficiency map and giving them the right kind of scaffolds to help them meet and exceed the grade level standard. Leaders in education should encourage teachers and students to explore and experiment with these new ways of learning. Trying things out, even if they don’t work at first, is an important part of getting comfortable with new technology and methods. Where It’s Happening: Singapore is adding AI learning tools to support greater customization of learning for all students and to augment teachers’ professional practice. In particular, an AI- enabled adaptive learning system uses machine learning to make customized learning recommendations for each student, based on how the student performs on activities and responds to questions while learning a concept. Launched in June 2023 to support Mathematics, the system is being continually expanded to include more topics, levels, and subjects and to include natural language feedback. South Korea is in the process of replacing print-based textbooks with digital content available through an AI-powered learning system, a move aimed at providing more equitable access to powerful learning for all students while reducing the heavy reliance on private schools and tutors. By 2025, students will access AI-powered content as well as personalized AI-enabled tutoring via their devices. This will allow teachers to attend to students’ social and emotional behaviors and offer instruction that is more centered around active learning according to the Ministry of Education. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 12 Near Term: Now—end of 2025 Teachers must be supported by organizational structures and incentives that reflect the modern reshaping of professions that drive the knowledge and creative economies. The way we work, especially in knowledge-based jobs—those involving the application of expertise and analytical skills to create, manage, or disseminate information—is changing with the advent of AI. And the work environment for teachers must change too, to keep up with new technologies and innovations. This is already having an impact on work organization and use of productivity tools within professions. Organizations are adopting an AI-fueled “copilot” approach to support knowledge professionals. These AI copilots can talk in everyday language, understand what we’re saying, and help finish tasks faster, including both routine and some creative tasks. They work by spotting patterns and simplifying complex tasks to make decisions easier. They can also make it easier to communicate with various stakeholders. The education profession has an opportunity to follow a similar path in order to realize the same efficiencies and elevate its attractiveness and competitiveness for prospective educators. As AI capabilities continue to evolve they can support teacher professional learning by summarizing new research, developing and refining instructional materials, and identifying areas for continued growth. For instance, an AI copilot could assist teachers working together through action research by framing their research questions and helping to analyze results. AI can act as a virtual coach, keeping teachers informed about subject content and provide immediate responses on teaching resources and strategies. Also, AI can automate repetitive tasks like reviewing lesson plans, learning resources, student assignments, and assessments for coherence and alignment to district guidelines. Across the Uited States, a lot of teachers are already self-organizing to use GenAI tools in this way, collaborating with colleagues to use the tools effectively, maximizing the value of outputs while minimizing the pitfalls. Additionally, there are initiatives underway to train GenAI outputs to be more domain specific, accurate, and trustworthy when applied to teaching. In a recent report from the International Monetary Fund (IMF), the authors note that...AI’s capabilities extend to cognitive functions, enabling it to process vast amounts of data, recognize patterns, and make decisions. As a result, even high-skill occupations, which were previously considered immune to automation because of their complexity and reliance on deep expertise, now face potential disruption. Jobs that require nuanced judgment, creative problem-solving, or intricate data interpretation—traditionally the domain of highly educated professionals—may now be augmented or even replaced by advanced AI algorithms… The Society for Human Resource Management (SHRM) in collaboration with The Burning Glass Institute recently issued a report in which they note the following One of the most visible ways GenAI will impact jobs is in the automation of repetitive professional tasks that require low levels of expertise or judgment. Simultaneously, GenAI may augment or transform other roles…GenAI can assist professionals by enhancing their capabilities, making them more effective or efficient rather than replacing them. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS 13 Near Term: Now—end of 2025 GenAI models need to be used to increasingly support leadership responsibilities. It is easy to focus entirely on how AI can impact teaching and learning, but AI can support school and district leaders as well. GenAI, coupled with more traditional predictive, rules-based machine learning (ML), is transforming management tasks within most knowledge industries, and education can benefit from these tools in the same way. GenAI can be used to support strategic thinking and planning, innovative problem solving, scenario- planning, and the development and delivery of communications to a wide variety of stakeholders.10 Classic predictive AI models can support functions such as resource allocation, scheduling, and data- driven decision making; provide practical guidance for categorizing upcoming events and learning agendas; or help anticipate future outcomes. Perhaps most important, while in a learning or piloting phase relating to the adoption of AI tools, system leaders should keep two key actions in mind. These points have been adapted from a recent report by World Economic Forum and the global Talent/Human Potential practice of Accenture)11: “Drive the adoption of LLMs [GenAI tools] by encouraging innovation and increasing awareness of how LLMs can benefit workers [the entire education community], including by automating routine tasks and augmenting human potential, resulting in more meaningful and fulfilling jobs.” “Ensure the adoption of transparent and inclusive LLM governance and frameworks. This approach necessitates inclusive and participatory decision-making processes in all aspects of LLMs governance and deployment, addressing not only efficacy but also bias, ethical implications and overall impact on job quality.” Many districts are finding ways to bring a range of stakeholders, including parents, businesses, and community members, together to learn about AI—its potential and risks—in order to be more transparent about decision making and policies. 10 Increasingly, generative AI tools are allowing data sets and documents that are specific and proprietary to end users to be 14 used to train foundation models with no risk of sharing confidential data. 11 Jobs of Tomorrow: Large Language Models and Jobs—A Business Toolkit, World Economic Forum, December 2023. Longer Term: 2026—2030 Education systems must set high expectations for students, focusing on the knowledge, skills, and dispositions necessary to succeed in a world marked by constant change and technological advancement, and develop performance assessments that measure them. While current state standards have served as a foundation for preparing young people to be college- and career-ready, they reflect a world that is receding into the past. To fully prepare for today and for the future, it’s essential to modernize a commonly held definition of what it means to be ready for college, career, and community and integrate knowledge, skills, and dispositions not currently accounted for in today’s standards. GenAI tools are now scoring as well as, or better than, top students on tests graded by machines. This means that the core knowledge and skills currently taught in schools is insufficient to prepare students for the future. Curriculum and assessments need to be built around the more complex, human-based reasoning that AI can’t do yet, ensuring that learning is cross-disciplinary and experiential. Additionally, students will need the ability to balance competing demands from multiple projects, the capacity to continually monitor their environment for new opportunities and engagements, a mindset of agility and agency for continual learning and growth (knowing that they will need to constantly evolve in terms of capabilities), and the collaborative behaviors needed to work in an environment of shared leadership. The post-secondary world is exploring the characteristics of human-centered work environments that use AI-powered technology and is creating profiles of the skills needed to flourish within these environments. One current view is that “the value of human capabilities that transcend specific skill sets and functional domains persists in ways that hard skills cannot, potentially making them more important than ever.” Human capabilities such as curiosity, empathy, divergent thinking, connected teaming, resilience, and social and emotional intelligence can be cultivated through experience and practice. That is precisely what schools need to be designed to do. For education systems to be learner-centered and future-ready, they need to be built around student learning goals that are holistic, focused on developing well-rounded students, and supported by student learning experiences and outcomes.12 12 2024 Global Human Capital Trends, Deloitte Insights, 2024. 15 Longer Term: 2026—2030 Education systems must be significantly modified to align to newly-defined high expectations, resulting in a dramatic change to how students acquire knowledge and skills, how they work together, how they apply their learning, and how they develop agency. A significant amount of investments in AI-powered education technologies are focused on improving current aligned instructional systems (curriculum, instruction, assessment, and teacher professional learning). In fact, curriculum and instruction that supports student proficiency as measured by machine-scored tests is likely, over time, to be largely automated through AI-powered adaptive learning environments embedded with personalized feedback and coaching. Preparing students for today and the future will require teachers to work with students in dramatically different ways, giving students more chances to use what they learn in real-life situations. This isn’t about adding things to the current system; it’s about reshaping the entire instructional system. Strong signals from emerging changes to work environments across a spectrum of economic sectors and job types have implications for how PK–12 learning environments are structured. One signal is the concept of “learning in the flow of work,” which implies that individuals can learn while completing tasks, rather than through isolated courses taken before beginning a task.13 This approach, which allows for learning when one has purpose, motivation and the opportunity to apply concepts, is reinforced by research over the past 30 years. Another signal is that workers will function more autonomously, collaborating with AI-powered copilots and other tools. Many organizations are hiring for mindset and contemporary workplace skills and will train employees on more technical skills, including STEM areas. Learning environments, whether in the classroom, hybrid, or fully online, will need to be highly learner-centric, and move from more structured in the earlier levels to less structured in the upper levels, with greater emphasis on student power-sharing and autonomy. The goal is that by the time students leave school, they will have had the opportunity to work more independently and across different pathways at the same time, and effectively manage their own learning. The environments that they will enter after completing initial schooling will demand that capacity.14 Explore More: Survey data from Gen Z (those born between 1997 and 2012) suggest that they are eager to work with cutting-edge technology and have opportunities to both learn and share;15 they will assess job offers based on the technology used by a prospective employer;16 they see themselves as creators and are eager to use AI and other tools;17 and they identify as entrepreneurs—they see the “side hustle” (ventures that they pursue outside of school and work) as a positive, not a negative.18 13 How to Help Your Team Learn in the Flow of Work, Helen Tupper and Sarah Ellis, Harvard Business Review, February 15, 2023. 16 14 International Labour Organization, World Economic Forum Presentation, January 2023. 15, 16 Gen Z—The Future Has Arrived, Dell Technologies, 2018. 17, 18 Freelance, side hustles, and gigs: Many more Americans have become independent workers, McKinsey & Company, 2022. Longer Term: 2026—2030 Teachers remain essential. Their role must evolve from helping students acquire knowledge and skills to becoming facilitators of inquiry and guiding students as they apply their learning to complex tasks. Teachers will need to support students with complex learning tasks that require the application of knowledge. However, the teacher may not be the primary source for core knowledge and skills. Instead, they will need to help students effectively use digital tools to develop their conceptual understanding and skills.19 Additionally, the teacher will need to effectively work with teams of students in helping them manage their own work, collaborate effectively, take responsibility for their own success, and keep track of their progress. This includes teaching students how to critically evaluate information provided by digital learning tools. Teacher preparation programs will need to focus on the art as well as the science of teaching—the human interactions between teachers and learners, both individually and in groups. Being a good teacher will require strong leadership and coaching skills just as much as knowing the subject matter—which will increasingly be supported by digital systems. Using AI tools as co-pilots for teachers can help schools create professional environments like those seen in top-performing education systems across the globe. These systems allow teachers time during the day for learning, collaboration, action research, and planning. By using AI and machine learning tools, schools can become more productive and creative in ways that are being demonstrated in other knowledge professions.20 This change could help school leaders and teachers to create environments in which “time and flexibility in the school day [enables] educators to collaborate with one another, further their professional learning goals, and exercise leadership to support the whole school, district, and system. Students, in turn, have dedicated time for 1:1 support, self-guided learning, and extracurricular activities that build a wide range of skills.”21 19 NCEE, through its NISL program, has redesigned its leadership development program from focusing on principals becoming instructional leaders to principals becoming architects of the system of capacity building for their teachers, who are expected to be (or become) experts in their disciplines. In an analogous way, the role of the teacher is increasingly evolving to that of architect of the learning environment for students. 20 Burning Glass Institute and Deloitte 17 21 Top Performers Reimagine the Teaching Profession, NCEE, 2023. Longer Term: 2026—2030 Leaders should create a culture of continual innovation and renewal, while ensuring a balance between introducing new ideas and maintaining stability within the system. New developments in GenAI, machine learning, and other technologies will likely continue to surprise and disrupt all sectors for the foreseeable future. However, following the first explosion of GenAI applications and tools, which directly accessed the foundation models (LLMs), there now is a maturity to the use of these new technologies, blending GenAI with machine learning analytics and more traditional non-AI technologies. This trajectory maps to the “learning, doing, optimizing” adoption curve22: initial learning (characterized by pilots and “trial-and-error” refinements) quickly moves to doing (characterized by adoption across networks and scaling of proven solutions) and finally matures to optimizing (characterized by a new normal, where standards, processes, and associated costs and benefits become the new normal). As part of this evolution, many companies in sectors outside of education are adopting an “AI Ops” approach in order to improve operational efficiencies (e.g., improve understanding of application and infrastructure dependencies), create new and differentiated service offerings, hire and retain talent, improve productivity across a remote workforce, and leverage data for analysis of trends that help them be proactive. Education leaders in this next phase need to take that “AI Ops” approach as they think about their entire system (see the NCEE Blueprint) and create the necessary structures, supports, and incentives to drive divergent thinking and innovation along with more structured operationalizing and optimization of new ideas. This aligns with the OECD idea of disciplined innovation23, but it’s not only getting students and teachers involved in co-creation but involves a continual scan of other sectors for ideas on how to integrate AI into workflow. Finally, building the human capacity of leaders becomes one of the most pivotal accelerators in driving PK–12 innovation. According to the Consortium for School Networking (CoSN), “[f]ostering leadership in others creates a culture of continuous improvement within schools. When leaders are visionary, adaptable, and well-informed, they set the stage for innovation to flourish and they create cultures where people want to engage and stay.”24 22 Rita McGrath and World Economic Forum adaptation 18 23 See Guideline 2, p. 7 in Opportunities, Guidelines and Guardrails for Effective and Equitable Use of AI in Education, OECD, 2023. 24 Driving K-12 Innovation 2024, p. 22, CoSN, 2024.  Appendix A: Developing Student Agency for Life and Work Note to the reader: this is still in development We want to ensure that all students, regardless of socioeconomic background, race, or cultural factors, have the same opportunities to flourish. This means that all students must be prepared to recognize and act on opportunities that contribute to their financial, physical, and emotional well- being. The education system should be designed to help build not just competences but the “muscles” to engage with opportunities and problems in an increasingly complex and changing world, and to take advantage of pathways to flourishing that impact them individually, across their communities, and within the institutions of which they are a part. We want, at a minimum, to be sure that ALL high school graduates are ready to succeed in the 2- and 4-year college programs or the onramps and apprenticeships to viable careers that will prepare them for modern work and further education. But we want far more than that for our students. We want them to learn how to lead and how to be good team members. We want them to set high standards for themselves and to be prepared to work hard to achieve them. We want them to do the right thing when it is not easy to do. We want them to take pleasure from being a contributing member of society and a friend and support to those around them. We want them to be tolerant and inclusive, tough and kind. Below are four categories of outcomes that education systems should consider as they prepare young people across the globe to thrive in a world that is increasingly disruptive—in terms of emerging opportunities and challenges, in the face of rapidly changing technologies, and in relation to external factors such as sustainability. These categories and the related descriptors are intentionally overlapping. They are not meant to be definitive or detailed. They are a guide to thoughtful conversations about transforming the school experience into one that builds students’ knowledge and muscle for an ever changing world and makes learning an active and joyful endeavor throughout their long life journey. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS i  Appendix A: Developing Student Agency for Life and Work Note to the reader: this is still in development Core Skills Students gain a broad and deep disciplinary knowledge base. They master academic standards comparable to those of the highest performing jurisdictions around the globe. Students’ mastery of core knowledge and skills increasingly requires the ability to work across disciplines and to apply learning in new contexts. Habits of Learning and Well-Being Students continually develop their agency for life and work. They embrace the power of learning and acquire skills that allow them to contribute to their own physical and mental health, overall well- being, and life satisfaction. They gain a sense of purpose about who they are and build an orientation toward a portfolio life—a balance of interests across a rich mix of work, family, community, and leisure. They contribute to their well-being and overall life satisfaction by building their capacity/ appetite to participate, and immersing themselves in activities that are uniquely human. These can be tied to culture such as the arts and humanities (e.g., engaging with works of art, playing and listening to music, reading a novel, engaging with historical works, writing a poem, studying a sacred or philosophical text, acting in a play), and/or connected to physical expressions of culture (e.g., dance, sports, athletic expression, etc). Contemporary Skills Students are well-equipped to deal with a world that is increasingly complex, fast changing, and uncertain, yielding an endless array of new opportunities and challenges. They develop skills connected with modern work success (e.g., communication, collaboration, creative thinking, ethical decision-making, agility, resilience, the ability to learn from and stare down failure). They apply both professional and common sense reasoning to respond to complexity in both work and daily life. They creatively and innovatively solve complex problems and use reasoning to convey their ideas clearly, often drawing on ideas from one discipline to inform another. They show a predilection for collaboration exhibiting the behaviors needed to work in an environment of shared leadership. They are able to work in teams and autonomously on their own. They demonstrate curiosity and compassion in their interaction with others in both their professional and personal lives. Community Skills Students develop skills that empower them to contribute beyond themselves (e.g., personal and civic responsibility, courage, respect, empathy, fairness). They seek out other perspectives and contexts, and actively build a view of the world that transcends their own geographic and virtual lens. They regularly think about their role within the context of a larger community, and the needs of the community in relation to their own. They weigh the perspectives and interests of others to inform decisions and actions, relating to what is fair or equitable, and their impact relative to the common good. They muster the courage to make choices that might not be popular but are deeply and authentically reflective of their informed point of view. FRAMEWORK FOR AI-POWERED LEARNING ENVIRONMENTS ii