Search for: All records

Automated methods are becoming increasingly used to support formative feedback on students’ science explanation writing. Most of this work addresses students’ responses to short answer questions. We investigate automated feedback on students’ science explanation essays, which discuss multiple ideas. Feedback is based on a rubric that identifies the main ideas students are prompted to include in explanatory essays about the physics of energy and mass. We have found that students revisions generally improve their essays. Here, we focus on two factors that affect the accuracy of the automated feedback. First, learned representations of the six main ideas in the rubric differ with respect to their distinctiveness from each other, and therefore the ability of automated methods to identify them in student essays. Second, sometimes a student’s statement lacks sufficient clarity for the automated tool to associate it more strongly with one of the main ideas above all others. 

With an increasing focus in STEM education on critical thinking skills, science writing plays an ever more important role. A recently published dataset of two sets of college level lab reports from an inquiry-based physics curriculum relies on analytic assessment rubrics that utilize multiple dimensions, specifying subject matter knowledge and general components of good explanations. Each analytic dimension is assessed on a 6-point scale, to provide detailed feedback to students that can help them improve their science writing skills. Manual assessment can be slow, and difficult to calibrate for consistency across all students in large enrollment courses with many sections. While much work exists on automated assessment of open-ended questions in STEM subjects, there has been far less work on long-form writing such as lab reports. We present an end-to-end neural architecture that has separate verifier and assessment modules, inspired by approaches to Open Domain Question Answering (OpenQA). VerAs first verifies whether a report contains any content relevant to a given rubric dimension, and if so, assesses the relevant sentences. On the lab reports, VerAs outperforms multiple baselines based on OpenQA systems or Automated Essay Scoring (AES). VerAs also performs well on an analytic rubric for middle school physics essays. 

Automated writing evaluation (AWE) systems automatically assess and provide students with feedback on their writing. Despite learning benefits, students may not effectively interpret and utilize AI-generated feedback, thereby not maximizing their learning outcomes. A closely related issue is the accuracy of the systems, that students may not understand, are not perfect. Our study investigates whether students differentially addressed false positive and false negative AI-generated feedback errors on their science essays. We found that students addressed nearly all the false negative feedback; however, they addressed less than one-fourth of the false positive feedback. The odds of addressing a false positive feedback was 99% lower than addressing a false negative feedback, representing significant missed opportunities for revision and learning. We discuss the implications of these findings in the context of students’ learning. 

Eighth grade students received automated feedback from PyrEval - an NLP tool - about their science essays. We examined essay quality change when revised. Regardless of prior physics knowledge, essay quality improved. Grounded in literature on AI explainability and trust in automated feedback, we also examined which PyrEval explanation predicted essay quality change. Essay quality improvement was predicted by high- and medium-accuracy feedback. 

The present study examined teachers’ conceptualization of the role of AI in addressing inequity. Grounded in speculative design and education, we examined eight secondary public teachers’ thinking about AI in teaching and learning that may go beyond present horizons. Data were collected from individual interviews. Findings suggest that not only equity consciousness but also present engagement in a context of inequities were crucial to future dreaming of AI that does not harm but improve equity. 

Helping students learn how to write is essential. However, students have few opportunities to develop this skill, since giving timely feedback is difficult for teachers. AI applications can provide quick feedback on students’ writing. But, ensuring accurate assessment can be challenging, since students’ writing quality can vary. We examined the impact of students’ writing quality on the error rate of our natural language processing (NLP) system when assessing scientific content in initial and revised design essays. We also explored whether aspects of writing quality were linked to the number of NLP errors. Despite finding that students’ revised essays were significantly different from their initial essays in a few ways, our NLP systems’ accuracy was similar. Further, our multiple regression analyses showed, overall, that students’ writing quality did not impact our NLP systems’ accuracy. This is promising in terms of ensuring students with different writing skills get similarly accurate feedback. 

In this paper, we present a case study of designing AI-human partnerships in a realworld context of science classrooms. We designed a classroom environment where AI technologies, teachers and peers worked synergistically to support students’ writing in science. In addition to an NLP algorithm to automatically assess students’ essays, we also designed (i) feedback that was easier for students to understand; (ii) participatory structures in the classroom focusing on reflection, peer review and discussion, and (iii) scaffolding by teachers to help students understand the feedback. Our results showed that students improved their written explanations, after receiving feedback and engaging in reflection activities. Our case study illustrates that Augmented Intelligence (USDoE, 2023), in which the strengths of AI complement the strengths of teachers and peers, while also overcoming the limitations of each, can provide multiple forms of support to foster learning and teaching. 

Building causal knowledge is critical to science learning and scientific explanations that require one to understand the how and why of a phenomenon. In the present study, we focused on writing about the how and why of a phenomenon. We used natural language processing (NLP) to provide automated feedback on middle school students’ writing about an underlying principle (the law of conservation of energy) and its related concepts. We report the role of understanding the underlying principle in writing based on NLP-generated feedback. 

Writing and revising scientific explanations helps students integrate disparate scientific ideas into a cohesive understanding of science. Natural language processing technologies can help assess students’ writing and give corresponding feedback, which supports their writing and revision of their scientific ideas. However, the feedback is not always helpful to students. Our study investigated 241 middle school students’ a) use of feedback, b) how it affected their revisions, and c) how these factors affected students’ writing improvement. We found that students made more content-related revisions when they used feedback. Making content-related revisions also assisted students in improving their writing. But students still found it difficult to make integrated revisions and did not use feedback often. Additional support to assist students to understand and use feedback, especially for students with limited science knowledge, is needed. 

Writing scientific explanations is a core practice in science. However, students find it difficult to write coherent scientific explanations. Additionally, teachers find it challenging to provide real-time feedback on students’ essays. In this study, we discuss how PyrEval, an NLP technology, was used to automatically assess students’ essays and provide feedback. We found that students explained more key ideas in their essays after the automated assessment and feedback. However, there were issues with the automated assessments as well as students’ understanding of the feedback and revising their essays.