*Personalized Learning: The Conversations We’re Not Having** * Data & Society Working Paper 07.22.2016 MONICA BULGER *Abstract* The pursuit of personalized education at a mass scale drives a number of current education technology initiatives that fall under the broad umbrella of personalized learning. The personalized learning space has seen a huge infusion of capital in recent years, with notable support from the Bill & Melinda Gates Foundation and a range of other investors, philanthropists, and technologists. Federal support for personalized learning has come through Obama’s Race to the Top program and, looking ahead, will come through the recently passed Every Student Succeeds Act’s provision for competency-based assessments. The story of personalized learning is a complex one in which education technology developers are applying startup logics like risk taking and testing in the field and marketing tactics like personalized content delivery and recommender systems. Their products are designed to meet the needs of states and administrators, who seek to improve efficiency and performance through formative testing and tailored learning modules, but they are often narrated differently to the public. The promise of technologically enabled personalized learning – that through technology we can democratize access to information and instruction, scale excellence across the socioeconomic spectrum, and do so with efficiency and rigor – is a familiar one to educators. And yet there are tensions between the promises of personalized learning, the state-of-its-art, and the practical realities of its implementation. As a buzzword symbolizing the potential of data use in education, personalized learning enjoys a broad definitional reach and is aligned by proponents with educational best practices like “student-centered instruction.” At its best, personalized learning represents a pedagogy of adaptation to students’ unique combination of goals, interests, and competencies. But many so-called personalized learning platforms are more responsive than adaptive, as they provide materials appropriate to students’ proficiency levels without moving beyond a pre-determined decision tree. They do this by analyzing student data such as age, gender, grade level, and test performance against idealized models based on data sets of students with similar demographics, then offering content recommendations directly to the student or to the teacher for further intervention. The widespread adoption of personalized learning platforms may lead to unintended consequences. For example, the focus on performance may negatively affect students’ psychological wellbeing by promoting individualist notions of achievement and undermining the social dimensions of learning. With the wealth of data that such systems generate, it becomes possible to track and report students’ progression, reinforcing a laser focus on numbers and performance metrics which may undermine other key facets of learning, while forcing teachers to support algorithmic logics rather than leverage their knowledge about education. In addition, data-driven personalized learning algorithms depend upon data points but there is a lack of transparency over what data points are collected, how they are used, and who is permitted access, leading to the recent sharp uptick in proposed national and state student data privacy legislation. Full PDF here: http://www.datasociety.net/pubs/ecl/PersonalizedLearning_primer_2016.pdf