Trustworthy Generative AI

We see lots of news reports of Generative AI tools, such as ChatGPT, making mistakes and producing inaccurate information. Many of these mistakes happen because humans use the tools in the wrong way – trying to solve unsuitable problems and not thinking about risk. Hallucination isn’t a bug, it’s a feature when you approach problems correctly. This course teaches techniques for determining if a problem fits Generative AI’s capabilities, framing problems to reduce risk, prompt engineering for trust, and appropriate human engagement in the process. Students learn concrete prompt designs, how to check outputs, how to use Generative AI for ideation and creation, ways to augment human skills, and more ethical, beneficial applications.