Overview

Cycle 2 of the FRG focuses on artificial intelligence (AI), one of HBKU’s focus areas. It prioritizes Generative AI (GenAI), which is rapidly transforming our world and has the potential to address some of humanity’s most pressing challenges, including sustainable development, precision medicine, and personalized education. However, GenAI also poses new challenges, such as the potential for bias, misuse, and unintended consequences.

HBKU researchers in multidisciplinary, cross-entity teams will focus on the following topics:

  • AI for Personalized Education
  • Safe, Responsible, and Ethical AI
  • AI for Science

Project 1: Advancing Qatar's AI Landscape: Developing an Arabic-Centric and Ethically-Aligned Large Language Model

This project aims to develop a robust, ethically-aligned Large Language Model (LLM) that embodies the Arabic and Islamic identity. Our approach will build on Arabic-centric LLMs that are pre-trained with Arabic language text, by customizing the alignment and fine-tuning phases to reflect the unique linguistic and cultural context of Qatar more accurately. This effort seeks to create a new standard in AI, focusing on technology that is not only cutting-edge but also culturally and ethically relevant.

The study will focus on four primary objectives:

  • Creating a dataset for multiplex ethics
  • Developing alignment methods compatible with multiplexity
  • Detecting attacks targeting human-value alignment of LLMs
  • Showcasing diverse application scenarios

Major outputs include:

  • Dataset for multiplex ethics
  • New alignment methods
  • Fine-tuned models for different applications
  • Publications
  • Teaching material for multiplex AI ethics

Team members

Project 2: AI-EDAPT: Artificial Intelligence for Educational Adaptation, Personalization, and Transformation

The project aims to advance the understanding and implementation of GenAI tools in higher education settings, focusing on enhancing personalized learning experiences. By developing a system that uses LLMs, such as LLaMA2 and ChatGPT, to monitor and analyze student interactions with GenAI tools, it seeks to improve teaching methods and learning outcomes and tailor educational practices to individual learning styles.

The study focuses on:

  • Developing and integrating an innovative system based on open-source LLM to monitor and analyze students' interactions with GenAI tools within the educational context of HBKU.
  • Evaluating and analyzing the impact of ChatGPT1 (e.g., GPT-4, LLaMa2) on a) course design and teaching activities, b) student learning and critical thinking, c) the relationship between student learning styles and their use of ChatGPT
  • Providing empirical evidence and insights into the effective use of ChatGPT as EduLLM in education, contributing to developing guidelines and best practices for its implementation.
  • Disseminating the findings and recommendations to the broader educational and academic communities through workshops, publications, and presentations.

Major outputs include:

  • Application to monitor student interactions with GenAI tools
  • Analysis reports on student interactions
  • Ethical guidelines and best practices
  • Publications

Team members