Generative AI Models and Prompts

This unit surveys major generative and specialized machine learning models (ChatGPT, DALL·E, CTRL, T2T, BERT, Tacotron, MuseNet), focusing on how prompts influence outputs, practical use cases, and current limitations. Students will examine ethical and societal implications and practice prompt design across text, image, and audio modalities to prepare for applied prompt engineering work in subsequent units.

Learning Objectives

• Analyze the architectures and functional roles of ChatGPT, DALL·E, CTRL, T2T, BERT, Tacotron, and MuseNet and describe how prompt structure influences each model’s output
• Compare strengths, weaknesses, and typical applications of text, image, and audio generative models using real-world examples
• Apply prompt-design techniques to create, test, and iteratively refine prompts for at least three model modalities (text, image, audio), documenting observed differences in outputs
• Evaluate ethical, legal, and safety concerns (including bias, copyright, and misuse) associated with deploying generative models and propose mitigation strategies for selected use cases