We introduce Generalized Instruction Tuning (called GLAN), a general and scalable method for instruction tuning of Large Language Models (LLMs). Unlike prior work that relies on seed examples or existing datasets to construct instruction tuning data, GLAN exclusively utilizes a pre-curated taxonomy of human knowledge and capabilities as input and generates large-scale synthetic instruction data across all disciplines. Specifically, inspired by the systematic structure in human education system, we build the taxonomy by decomposing human knowledge and capabilities to various fields, sub-fields and ultimately, distinct disciplines semi-automatically, facilitated by LLMs. Subsequently, we generate a comprehensive list of subjects for every discipline and proceed to design a syllabus tailored to each subject, again utilizing LLMs. With the fine-grained key concepts detailed in every class session of the syllabus, we are able to generate diverse instructions with a broad coverage across the entire spectrum of human knowledge and skills. Extensive experiments on large language models (e.g., Mistral) demonstrate that GLAN excels in multiple dimensions from mathematical reasoning, coding, academic exams, logical reasoning to general instruction following without using task-specific training data of these tasks. In addition, GLAN allows for easy customization and new fields or skills can be added by simply incorporating a new node into our taxonomy.

Abstract: Recent work has explored how complementary strengths of humans and artificial intelligence (AI) systems might be productively combined. However, successful forms of human-AI partnership have rarely been demonstrated in real-world settings. We present the iterative design and evaluation of Lumilo, smart glasses that help teachers help their students in AI-supported classrooms by presenting real-time analytics about students' learning, metacognition, and behavior. Results from a field study conducted in K-12 classrooms indicate that students learn more when teachers and AI tutors work together during class. We discuss implications for the design of human-AI partnerships, arguing for participatory approaches to research in this area, and for principled approaches to studying human-AI decision-making in real-world contexts. Artificial intelligence (AI) systems are increasingly used to support human work in deeply social contexts such as education, healthcare, social work, and criminal justice. In these contexts, AI can automate routine parts of practitioners' work, while freeing up their time for activities they find more meaningful (17, 28, 39).

Robots can take just three hours to successfully learn techniques which can be used to support teachers in a classroom environment, according to new research. The study, published in Science Robotics, saw a robot being programmed to progressively learn autonomous behaviour from human demonstrations and guidance. A human teacher controlled the robot, teaching it how to help young pupils in an educational activity, and it was then able to support the children in the same activity autonomously. The advice it subsequently provided was shown to be consistent with that offered by the teacher. Researchers say the technique could have a number of benefits to teachers, as they face increasing demands on their time, and could be positive for pupils, with research previously showing that using robots alongside teachers in the classroom can have benefits for their education.

This course is part of a certificate program. You can enroll in this course on a space available basis even though you are not a certificate student. Courses taken when you are not a certificate student don't automatically count toward earning a certificate. You will pay a course fee when you are notified of your eligibility to enroll. We will let you know whether you are accepted or not accepted into the course before the first class session.