Conference schedule overview

Second, it can refer to the learning process of machines, notably to ‘machine learning’ as one of the most promising techniques of artificial intelligence. I will call this the Herbert Simon approach, even though Simon was quite sceptical of machine learning. Third, ‘learning as a machine’ can refer to the fact that human beings increasingly live in a world saturated with data-driven applications that are more or less capable of machine learning. Since this will require human beings to anticipate how their intelligent environment learns, I will argue that – to some extent – humans will engage in ‘learning as if a machine’. In my keynote I will investigate what this could mean in terms of human liberty and human dignity – two key terms in the privacy debate – and explain how it relates to the employment of learning analytics to aid human learning processes. This will include a discussion of legal protection by design in the context of learning analytics, notably providing students with profile transparency while protecting their fundamental right to data protection. For an example of profile transparency in another context, see http://www.usemp-project.eu. Read less…

First the good news. LA is probably the first in the long line of innovations and promises based upon them that can possibly achieve many of the educational futures that people are hoping for such as adaptivity, differentiation, tailor-made instruction, and so forth. And now the bad news. While most of the others failed and in their failure were innocuous (if you consider wasting time and money to be innocuous), LA has the potential to also do harm in a number of ways. This keynote will try to put both the possible utopian futures and their dystopian counterparts on the research and praxis agendas. Read less…

Some are familiar, but others are quite new to the field itself. Key insights that have evolved over the years have been (1) viewing the problem as one of statistical inference, (2) building models that suited the inferential problem as cast in then-current psychological theory, informed by then-current forms of data, and (3) developing methods to define and operationalize properties such as reliability, validity, comparability, generalizability, and fairness – not just measurement principles, Messick (1994) reminds us, but “social values that have meaning and force outside of measurement wherever evaluative judgments and decisions are made” (p. 13). Rapid advances in technology and sociocognitive branches of psychology have sparked a renaissance in the field, as psychometricians work across disciplines to disentangle key insights from the particulars of historical models, forms of data, and psychological perspectives. Current work seeks to synthesize hard-won psychometric understandings with computational data analysis methods; results from the learning sciences, domain-based research, and cognitive and sociocultural psychology; and simulation, game, and immersive environments. Several of the challenges we are tackling connect with those faced in learning analytics. I will note connections where I see them, with an anticipation of mutual benefit. Read less…