Sorting molecules with DNA robots: Single-stranded DNA robots can move over the surface of a DNA origami sheet and sort molecular cargoes. Thubagere et al. developed a simple algorithm for recognizing two types of molecular cargoes and their drop-off destinations on the surface (see the Perspective by Reif). The DNA robot, which has three modular functional domains, repeatedly picks up the two types of molecules and then places them at their target destinations. No additional power is required because the DNA robot does this by random walking across the origami surface. Science , this issue p. eaan6558 ; see also p. 1095
4 October 2022 in International Journal of Artificial Intelligence in Education
In this paper, I argue that the fields of artificial intelligence (AI) and education have been deeply intertwined since the early days of AI. Specifically, I show that many of the early pioneers of AI were cognitive scientists who also made pioneering and impactful contributions to the field of education. These researchers saw AI as a tool for thinking about human learning and used their understanding of how people learn to further AI. Furthermore, I trace two distinct approaches to thinking about cognition and learning that pervade the early histories of AI and education. Despite their differences, researchers from both strands were united in their quest to simultaneously understand and improve human and machine cognition. Today, this perspective is neither prevalent in AI nor the learning sciences. I conclude with some thoughts on how the artificial intelligence in education and learning sciences communities might reinvigorate this lost perspective.
The public release and surprising capacity of ChatGPT has brought AI-enabled text generation into the forefront for educators and academics. ChatGPT and similar text generation tools raise numerous questions for educational practitioners, policymakers, and researchers. We begin by first describing what large language models are and how they function, and then situate them in the history of technology’s complex interrelationship with literacy, cognition, and education. Finally, we discuss implications for the field.
14 November 2019 in International Journal of Artificial Intelligence in Education
Since the 1960s, researchers have been trying to optimize the sequencing of instructional activities using the tools of reinforcement learning (RL) and sequential decision making under uncertainty. Many researchers have realized that reinforcement learning provides a natural framework for optimal instructional sequencing given a particular model of student learning, and excitement towards this area of research is as alive now as it was over fifty years ago. But does RL actually help students learn? If so, when and where might we expect it to be most helpful? To help answer these questions, we review the variety of attempts to use RL for instructional sequencing. First, we present a historical narrative of this research area. We identify three waves of research, which gives us a sense of the various communities of researchers that have been interested in this problem and where the field is going. Second, we review all of the empirical research that has compared RL-induced instructional policies to baseline methods of sequencing. We find that over half of the studies found that RL-induced policies significantly outperform baselines. Moreover, we identify five clusters of studies with different characteristics and varying levels of success in using RL to help students learn. We find that reinforcement learning has been most successful in cases where it has been constrained with ideas and theories from cognitive psychology and the learning sciences. However, given that our theories and models are limited, we also find that it has been useful to complement this approach with running more robust offline analyses that do not rely heavily on the assumptions of one particular model. Given that many researchers are turning to deep reinforcement learning and big data to tackle instructional sequencing, we believe keeping these best practices in mind can help guide the way to the reward in using RL for instructional sequencing.
In addition to providing a set of techniques to analyze educational data, I claim that data science as a field can provide broader insights to education research. In particular, I show how the bias-variance tradeoff from machine learning can be formally generalized to be applicable to several prominent educational debates, including debates around learning theories (cognitivist vs. situativist and constructivist theories) and pedagogy (direct instruction vs. discovery learning). I then look to see how various data science techniques that have been proposed to navigate the bias-variance tradeoff can yield insights for productively navigating these educational debates going forward.