Practical Strategies for Ethical Use of Generative AI in Scientific Research Practices
Practical strategies are needed to bridge the gap between abstract ethical principles and day-to-day use of generative AI tools in scientific research practices.
Automating Computational Reproducibility: Evaluating AI Agents on a Benchmark for Reproducing Published Research
Automating the computational reproducibility of published research is a crucial yet challenging task that can significantly improve the credibility of scientific findings.
Effective Real-Time Monitoring of Online Decision-Making Algorithms in Digital Health Intervention Trials
Effective real-time monitoring systems are essential to safeguard participants and ensure data quality when using online decision-making algorithms in digital health interventions.
Δ-OPE: Off-Policy Estimation with Pairs of Recommendation Policies
The difference between the value of a target recommendation policy and a production policy can often be estimated with significantly reduced variance compared to estimating the value of each policy individually.
Opportunities and Challenges for AI-Driven Replication Prediction to Improve Confidence in Published Research: Perspectives from India
Researchers in India see value in AI-driven tools for assessing the replicability of published findings, but emphasize the need for transparency, explainability, and human-AI collaboration to build trust in such systems.
NovAScore: An Automated Metric for Evaluating Document-Level Novelty
NOVASCORE is an automated metric that evaluates the novelty of a target document by aggregating the novelty and salience scores of its atomic content units, providing detailed analysis and strong correlation with human judgments of novelty.
Quantitative Analysis of Language Model Usage and Trust in Academia: Insights from a Comprehensive Survey
The majority of students, staff, and faculty in academia actively use language models, and increased usage is positively correlated with higher levels of trust in these tools. Fact-checking is perceived as the most critical issue to prioritize for the responsible development of language models.
The Impact of Large Language Models on Learning Outcomes in Coding Classes
Having access to large language models like ChatGPT can have both positive and negative effects on student learning outcomes in programming courses, depending on how students use the technology.
An Unsupervised Dialogue Topic Segmentation Model Based on Utterance Rewriting
This study proposes a novel unsupervised dialogue topic segmentation method that combines Utterance Rewriting (UR) technique with an unsupervised learning algorithm to efficiently utilize the useful cues in unlabeled dialogues by rewriting the dialogues in order to recover the co-referents and omitted words.
Contestable AI Framework for Interactive Feedback in Evaluating Student Essays
A Contestable AI Empowered LLM Framework (CAELF) that integrates computational argumentation to automate interactive feedback, enhancing the reasoning and interaction capabilities of language models in educational settings.
© 2024 by Linnk AI