Agenda

Contents

• What are large language models (LLMs)? How do they generate text?
• What tasks can LLMs perform? What are their limitations?
• What are effective prompting techniques?

Slides

Prompt Labor: Basics

Prompt Labor: Vertiefung

Activities

1. Prompting techniques

2. Confabulations and prompting

3. Prompt Labor: Vertiefung

Take-home messages

• Explore LLMs firsthand to understand their strengths and weaknesses.
• Combine domain knowledge with an understanding of how LLMs work, and effective prompting strategies.
• Integrate LLMs into teaching to foster AI literacy among students.
• Critically evaluate an LLM’s output. They are language models, not knowledge bases.
• Keep a human in the loop.

Learning outcomes

After this workshop, you will be able to:

1. Explain what large language models and conversational agents, such as ChatGPT, can be used for, and what they shouldn’t be used for.
2. Create effective prompts for LLMs.

Instructors

• Andrew Ellis, Virtual Academy at the Bern University of Applied Sciences:

  Andrew, a data scientist at the Virtual Academy, explores the convergence of language, thought, and AI. He teaches and researches generative AI’s role in education at BFH, focusing on human interactions with large language models and the impact of AI learning tools on educational outcomes. Andrew holds a PhD in cognitive psychology from the University of Bern, where he studied mental imagery and perception.

• Kaspar Kaufmann, Virtual Academy at the Bern University of Applied Sciences:

  As a researcher at the Virtual Academy, Kaspar aims to promote digital competencies among teachers and learners at BFH. He is passionate about fostering a community of confident, critical and responsible users of digital technologies for education, work and civic participation.