What Can Meeting Leaders Learn from AI Prompt Engineers?

Meetings are the backbone of collaboration and decision-making within an organization. They serve to convene the group or team, create shared knowledge and experience, and ignite forward momentum of work. However, without the proper knowledge, skills, and preparation, meeting leaders can struggle with facilitating effective and efficient meetings.

In a similar way, Artificial Intelligence (AI) is a tool that can aid collaboration, decision-making, and create shared knowledge. Large Language Models (LLM’s) like ChatGPT, PaLM 2, and DALL-E-2 can summarize, classify, translate, answer questions, and generate images. However, without clear and rigorous training, these AI systems can be prone to providing incorrect information or not knowing how to interpret an input. In other words, AI needs a translator and to be trained.

AI prompt engineers are tasked with ‘training’ these generative AI systems and have been a recent hot topic in the media, perhaps due to the promise of making a $300k/year salary without needing a computer science degree. Prompt Engineering encompasses the discipline of creating prompts, analyzing the outputs, and making refinements to improve the capacity of LLMs. These feedback loops help to build a system and sheds light on ways we can learn, create, design, and evolve.

So what insights can we glean from prompt engineering? In this blog, I’ll outline specifically a few things meeting leaders and facilitators can apply in your next meeting.

Understand the Context and Objectives
Prompt engineers must first understand the data and context deeply. They understand how to tailor prompts to a specific AI system, use case, and outcome they are aiming to achieve. From a technical perspective, they understand the desired response length, the balance between creativity and plausibility, and the desired heading, output structure, and format.

Understanding the context and end goal allows the prompt engineer to come up with a helpful way to frame their prompt. Prompts are made up of any of the following elements: instruction (what is the specific task), context (other information), input (the question), and output (the desired output format).

Clear prompt instructions involve clear commands such as: write, classify, summarize, translate.
____________________________________________________________________________________
Prompt example:
“Write a 150-word summary about the book The Catcher in the Rye. Explain it to me like I am 5 years old. Use bullet points. Use reliable sources and cite those sources.”

Output from ChatGPT:
"The Catcher in the Rye" is a copyrighted work by J.D. Salinger, and while I can't provide a direct citation, I can offer a simplified explanation suitable for a young audience based on its general themes.

• The Story: It's a book about a boy named Holden who is unhappy and feels alone.
• Adventures in New York: Holden leaves school and goes to a big city called New York. There, he meets different people and does fun and sometimes strange things.
• Not Trusting Grown-Ups: Holden doesn’t think that adults are always honest or kind. He wishes to protect kids from growing up too fast.
• Feeling Lost: He talks about feeling confused and like he doesn’t belong in the world.
• Growing Up Can Be Tough: The book helps us understand that growing up can sometimes be hard and that it's okay to feel sad or alone sometimes.

This summary is based on the general themes and ideas commonly discussed in analyses and critiques of "The Catcher in the Rye."
_____________________________________________________________________________________

In a similar way, meeting leaders and facilitators should deeply understand the meeting purpose, the objectives, the meeting duration, and consider what would be a ‘win’ for meeting participants. This helps to create powerful Desired Outcome statements and agendas, which help guide the group towards intended meeting outputs.

Use Concise and Unambiguous Language.
Prompt engineers must have a deep understanding of linguistics – syntax, pragmatics, semantics. They understand that language is a symbolic system and that words are connected to each other. The use clear, concise, and unambiguous language.

Did you know that words are often stored in our brain based on their relationship to other words? {here is a brief activity that demonstrates how your mind stores words}. How is the structure of language formed? How are words and phrases arranged? How is meaning derived and implied? How do people use language to communicate and understand each other?

An example of pragmatics in action:
“Can you pass the salt?”
Literal meaning: are you able to do this task?
Pragmatic meaning: will you pass me the salt?

In a similar way, meeting leaders and facilitators can use language that is clear and concise. Avoid acronyms or ‘inside language’ that only a few in the group may understand. Ask straightforward questions. When leading a group through a conversation, take on one item or subject at a time. Get alignment on the problem before you go to solution space. When capturing action items, be specific. Who is the task owner, the task, and the specific due date?

Give Examples, then Refine.
Prompt engineers can refine AI models by providing examples of what a successful output would look like. Instead of simply asking the AI a question or providing a Yes/No response, Prompt Engineers work to refine the model by providing examples.

Prompt Example:
Q: What is the capital of France?
A: Paris
Q: What is the capital of Germany?
A: ?

In the above example, the AI is then being trained to answer the question in the desired format (Berlin).

Prompt engineers also use a technique called Chain-of-Thought prompting (CoT). This enables the LLM to have complex reasoning capabilities through intermediate reasoning steps. In a way, it breaks down the complexity into smaller, bite sized steps and guides the model using an example on how to solve such problem. The prompt engineer can even use the phrase “let’s go step by step” to elicit the model to elaborate step-by-step how they came up with the output.

In a similar way, Meeting leaders and facilitators can clearly give examples of the meeting process and take each element of the meeting step-by-step. We know from our own original research that over 90% of meeting problems are process-related, so providing ‘process commercials’ throughout a meeting can serve to educate the group about the flow or the conversational or information management approach.

Meeting Process Commercial Example:
For the next five minutes as a group, we are going to generate a list of our top strategic priorities as a company. I’ll first invite one or two people to come off mute to speak verbally (give example). Later, I’ll invite everyone to provide your input via Zoom Chat. John – our scribe for today – will capture all these ideas (verbal and written) into our meeting notes.

Take a Systems Approach
The best prompt engineers take a systems approach to improving training. They create robust conversational frameworks and templates in which to build, test, and scale AI systems. Prompt engineering involves using processes that are reusable and can adapt to new situations. They focus on building models of discourse versus simply a response or answer.

In a similar manner, Meeting Leaders can improve the effectiveness and efficiency of their meetings by taking a systems approach. Ask for feedback at the end of each meeting (what worked/what would we change) to create a continuous feedback loop. Have clear protocols across the organization for how meetings are set up or conducted.

The most effective meeting leaders model the Interaction Method: they are collaborative, strategic, receptive and flexible. They are most powerful and collaborative when they design and facilitate conversations that unlock rich discourse, relationships, and teamwork versus simply capturing an answer.

This helps to transform a single meeting into a series of meetings or an entire system that can drive organizational change. As more meeting leaders learn and practice the Interaction Method, meetings get more efficient and effective, leading to faster decision making, improved collaboration, and increased satisfaction across results, process, and relationship dimensions.

Are you a meeting leader looking to up your facilitation game? Check out our popular program Essential Facilitation^® and how it can take your facilitation skills to the next level.

*Image generated by DALL-E