AI Ethics & Responsible AI

This tutorial introduces Neuro-Symbolic AI as a method to enhance Large Language Models (LL Ms). It focuses on making LL Ms more robust, explainable, and instructable by combining symbolic knowledge structures with statistical learning techniques. The goal is to address the limitations of black-box LL Ms, particularly in terms of transparency and domain-specific protocol understanding.

This course explores the intersection of AI, ethics, and psychology, including topics like emotional AI and the psychological impact of human-machine interaction.

This course explores the best practices for responsible AI implementation in medicine. It is part of a specialized set of programs designed to train healthcare professionals to effectively use AI-driven tools, covering topics like bias in AI models, explainability, and the importance of human oversight.

This University of Michigan course explores the ethical considerations in data science, including fairness, accountability, and transparency, which are deeply connected to statistical concepts of bias and variance.

This course delves into the societal impact of facial recognition technology, which is increasingly a part of our daily lives. It covers the technical and design issues that can lead to biased performance across different societal groups and emphasizes the need for responsible deployment to mitigate these inherent biases.

A presentation exploring the state-of-the-art applications of AI in medical imaging through the lens of Responsible AI. It addresses critical ethical considerations such as bias mitigation, transparency, accountability, and data privacy, along with regulatory and implementation challenges.

Dive into the cutting-edge world of Direct Preference Optimization (DPO) and Large Language Model Alignment with this comprehensive course designed to equip you with the skills to leverage the LLaMA3 8-billion parameter model and Hugging Face's Transformer Reinforcement Learning (TRL). Using the powerful Google Colab platform, you will get hands-on experience with real-world applications, starting with the Intel Orca DPO dataset and incorporating advanced techniques like Low-Rank Adaptation (LoRA).Throughout this course, you will:Learn to set up and utilize the LLaMA3 model within Google Colab, ensuring a smooth and efficient workflow.Explore the capabilities of Hugging Face’s TRL framework to conduct sophisticated DPO tasks, enhancing your understanding of how language models can be fine-tuned to optimize for specific user preferences.Implement Low-Rank Adaptation (LoRA) to modify pre-trained models efficiently, allowing for quick adaptations without the need to retrain the entire model, a crucial skill for real-world applications.Train on the Intel Orca DPO dataset to understand the intricacies of preference data and how to manipulate models to align with these insights.Extend your learning by applying these techniques to your own datasets. This flexibility allows you to explore various sectors and data types, making your expertise applicable across multiple industries.Master state-of-the-art techniques that prepare you for advancements in AI and machine learning, ensuring you stay ahead in the field.This course is perfect for data scientists, AI researchers, and anyone keen on harnessing the power of large language models for preference-based machine learning tasks. Whether you're looking to improve product recommendations, customize user experiences, or drive decision-making processes, the skills you acquire here will be invaluable.Join us to transform your theoretical knowledge int