How can you use AI to write scenarios for learning? I am not an expert in AI by any stretch, but I have been experimenting with ChatGPT, Bard, and other AI tools. So far, I have found these tools helpful in generating ideas, writing first drafts, and summarizing. They work better for general knowledge tasks than really specific topics unless I provide more details to them, which makes sense.

This post isn’t going to give you “5 magical prompts to instantly write scenarios for you” or anything like that. Instead, this is a “working out loud” post where I’ll share some prompts I have used. I’ll also share my thoughts and how I refine the AI-written drafts for this example scenario. I’m using the free ChatGPT 3.5 and Bard for these examples. This will show the first part of writing a scenario, showing how I used AI to generate ideas, an opening scene, and two decision points. I’ll continue the scenario in my next blog post.

The image above was generated using Canva’s free AI image generation tool.

Topic: Workplace conflict

I’m using workplace conflict as my general topic here because it’s one where LLMs (large language models) like ChatGPT and Bard can be successful. I expect the tools to have plenty of source material to draw on. Plus, this isn’t a real project for a client, just a demo, so I’m less worried about accuracy.

For real workplace training projects, everything you generate with AI should be reviewed by SMEs for accuracy. Use the drafts, but assume there will be errors that need correcting. Be especially cautious using AI with high risk topics like safety or health care where the potential consequences of inaccuracy are significant.

Generating ideas

I used this prompt to generate some ideas for a scenario on workplace conflict.

“Give me 5 ideas for stories about workplace conflict. For each idea, tell me the names of 2 or more characters and the problem they are having.”

ChatGPT’s story ideas

Sure, here are 5 ideas for workplace conflict stories:

“The Meeting from Hell”: Linda and John are coworkers in the same department. They have very different opinions on how to approach a new project and clash during a team meeting. The situation escalates quickly, and they end up arguing in front of everyone, making the rest of the team uncomfortable and unproductive.

“The Backstabbing Coworker”: Sarah and Mike work in different departments but frequently collaborate on projects. One day, Sarah realizes that Mike has been taking credit for her work and presenting it to their boss as his own. Sarah confronts Mike, but he denies everything, making her look paranoid and unprofessional.

“The Boss from Hell”: Emily works for a boss, Mark, who is known for being a micromanager and constantly criticizing his employees. Emily tries her best to ignore his behavior, but one day, Mark publicly humiliates her in front of her colleagues, causing her to lose confidence and motivation.

“The Office Romance Gone Wrong”: Tom and Rachel work in the same office and start dating, but their relationship quickly turns toxic. They argue constantly and involve their colleagues in their drama, creating a tense and uncomfortable work environment.

“The New Hire with Attitude”: Jason is a new hire who comes into the company with a lot of confidence and a “know-it-all” attitude. He clashes with his coworkers, especially Jane, who has been with the company for years and knows the ins and outs of the business. Jason dismisses her ideas and makes her feel undervalued, causing tension and conflict between them.

Bard’s story ideas

Here are 5 ideas for stories about workplace conflict:

The Power Struggle: Two managers are vying for the same promotion, and the tension between them is palpable. The characters are John and Mary, and the problem is that they both want the same thing, and they’re not willing to back down.

The Unwanted Intern: A new intern is assigned to a team, and she quickly rubs everyone the wrong way. The characters are Sarah and the intern, and the problem is that the intern is inexperienced and arrogant, and she doesn’t seem to be willing to learn.

The Gossipers: A group of employees are constantly gossiping about each other, and it’s starting to create a hostile work environment. The characters are the gossipers and the targets of their gossip, and the problem is that the gossip is causing a lot of tension and stress.

The Micromanager: A manager is constantly micromanaging her employees, and it’s driving them crazy. The characters are the manager and her employees, and the problem is that the manager is not giving her employees any autonomy, and they’re starting to feel like they’re not trusted.

The Bullying: An employee is being bullied by her co-workers, and she’s not sure what to do. The characters are the bully and the victim, and the problem is that the bully is making the victim feel uncomfortable and unsafe at work.

Comparing the responses

ChatGPT provided names of characters. Bard interpreted my request for characters as meaning just roles for the characters; it only named one character. ChatGPT’s names all sound very white and basic though, so I would generally replace them anyway.

I like “The New Hire with Attitude,” so I’ll use that as the basis for a scenario. I’m going to change the names to Oliver and Rita though, so they’re a little less common. For simplicity, I’ll work just with ChatGPT moving forward in this example. In practice, I find that sometimes I get better results with one or the other, so it’s worth trying the same prompt in both.

Refine the prompt for the opening scene

Based on that concept, I want an opening scene for the scenario to set up the conflict. For this example, I’m working with the assumption that I’ll create some animated videos for this scenario. In my experiments so far, treating it as a video script sometimes gets better results than simply asking for dialogue between characters.

First opening scene prompt attempt

Write a 30-second script for a scene with the title “The New Hire with Attitude.” The video has two characters, Oliver and Rita. Oliver is a new hire who comes into the company with a lot of confidence and a “know-it-all” attitude. He clashes with his coworkers, especially Rita, who has been with the company for years and knows the ins and outs of the business. Oliver dismisses her ideas and makes her feel undervalued, causing tension and conflict between them. The script should include dialogue between Oliver and Rita as the emotional tension increases between them. The scene should end with the conflict unresolved.

With that prompt, ChatGPT gave me something with multiple locations with a gap of several days between. That’s not going to work for this scenario opening, so I refined my prompt. I made the request shorter and specified that I want a single conversation.

2nd opening scene prompt attempt

Write a 20-second script for a video with the title “The New Hire with Attitude.” The video has two characters, Oliver and Rita. Oliver is a new hire who comes into the company with a lot of confidence and a “know-it-all” attitude. He clashes with his coworkers, especially Rita, who has been with the company for years and knows the ins and outs of the business. Oliver dismisses her ideas and makes her feel undervalued, causing tension and conflict between them. The script should include dialogue between Oliver and Rita in a single conversation as they argue about a process. The scene should end as the argument becomes heated, with the conflict unresolved.

ChatGPT’s draft opening scene

Here’s ChatGPT’s response. This feels workable as the intro to a scenario in setting up a problem to solve. It starts right with some action, without wasting too much time on fluff and exposition.

Title: The New Hire with Attitude

INT. OFFICE – DAY

Oliver and Rita are standing near the coffee machine.

OLIVER
(smugly)
Rita, I’ve been analyzing our workflow, and it’s clear we need a major overhaul. I’ve got some revolutionary ideas.

RITA
(skeptical)
Oh, really? We’ve been running smoothly for years. What’s so revolutionary?

OLIVER
(condescending)
Well, your way is old-fashioned. We should automate the process and streamline everything.

Rita’s frustration grows evident.

RITA
(defensively)
My way has proven effective, Oliver. We can’t just abandon it for untested ideas.

OLIVER
(rolls eyes)
Times change, Rita. It’s time to step into the future.

RITA
(angry)
Future? Your ego is blinding you. My experience matters, and you need to respect that.

The argument becomes heated, and they both storm off in opposite directions, leaving the conflict unresolved.

FADE OUT.

Write a decision point and choices

Now that I have an introduction to the scenario, I need a decision point: a question with choices where learners need to decide what to do next.

My initial attempt at a prompt didn’t quite work; it tried to resolve the whole conflict in one step rather than giving me a single step that I could use. But the overall approaches in the response made sense, at least at a high level. I don’t want so much instructional feedback in my scenario; I want to continue this with more conversation.

Decision point and choices prompt

You are an instructional designer. You are creating a training practice exercise for managers on how to address workplace conflict. This practice exercise will be an interactive scenario in the style of a Choose Your Own Adventure story. Continuing from the video script you just generated, provide 3 choices of actions Rita and Oliver’s manager could take after their argument. The three choices should be possible answers to the question, “As Rita and Oliver’s manager, what do you do next?” Indicate which choice is the best choice.

ChatGPT’s first decision point

As Rita and Oliver’s manager, what do you do next?

Choice 1: Ignore the Conflict

You decide to ignore the conflict, hoping it will resolve itself over time. You figure that they are both adults and should be able to work it out on their own. However, this choice could lead to the tension between Rita and Oliver escalating further, potentially affecting the team’s dynamics and overall productivity.

Choice 2: Address the Conflict Privately

You choose to address the conflict privately by scheduling separate meetings with both Rita and Oliver. During these meetings, you actively listen to their perspectives, acknowledge their feelings, and encourage open communication. You help them understand the importance of collaboration and teamwork while emphasizing the value of their individual contributions. This choice is a step in the right direction as it allows both employees to express their concerns and provides an opportunity for resolution.

Choice 3 (Best Choice): Facilitate a Mediation Session

You opt for a more proactive approach and decide to facilitate a mediation session between Rita and Oliver. In this session, you act as a neutral mediator, guiding the conversation while allowing both parties to express themselves fully. Through active listening and effective communication techniques, you help them understand each other’s viewpoints and identify common ground. This approach fosters understanding, empathy, and collaboration, creating a potential pathway to resolve the conflict amicably and strengthen their working relationship.

Best Choice Summary:

Facilitating a mediation session is the best choice as it promotes open communication and allows Rita and Oliver to work through their differences with your support. By taking this approach, you can foster a more positive and productive work environment while ensuring that conflicts are addressed constructively.

Revised decision point 1

Based on ChatGPT’s draft, I’m going to revise the answers.

As Rita and Oliver’s manager, what do you do next?

Give them space to resolve the conflict on their own.

Talk to Rita and Oliver separately and privately.

Schedule a meeting to mediate with both employees.

Second decision point

I usually write the ideal path first, the path where the learner makes all the best decisions. That’s what I’ll do here as well. Generating plausible distractors is one of the most challenging parts of writing scenarios, so that’s what I want to try to get from ChatGPT.

Decision point prompt

Continuing with this interactive scenario, write a multiple choice question with 3 options on how to start the conversation for a mediation session. Indicate which answer is the best option.

This prompt worked a little better than my previous one. Telling it I want a multiple choice question got the format closer to what I want. The best answer is much longer than the wrong answers, but it’s not bad as a first draft. I also note that it says to create ground rules, but it jumps right into hearing perspectives instead of setting those rules. That’s the kind of error I think we have to watch for in these drafts.

ChatGPT’s decision point draft

As you begin the mediation session between Rita and Oliver, how do you start the conversation?

Option 1: “Alright, let’s get this over with. Rita, you start by stating your grievances.”

Option 2: “I called you both here today to sort out your issues. Oliver, you can begin by explaining why you think your approach is better.”

Option 3 (Best Option): “I appreciate both of you being here today. Let’s take a moment to set some ground rules and create a safe space for open communication. Rita, would you like to share your perspective first, or shall we start with Oliver?”

Explanation:

Option 3 is the best option because it acknowledges the importance of creating a safe and respectful environment for the mediation session. By expressing gratitude for their presence and proposing ground rules, it sets a positive tone for the conversation. Giving Rita and Oliver the choice of who wants to speak first allows them to feel heard and respected, encouraging them to share their perspectives more openly. This approach fosters a constructive dialogue and increases the likelihood of finding a resolution to the conflict.

Revised decision point

In my revision, I pause after saying you want to create ground rules.

As you begin the mediation session between Rita and Oliver, how do you start the conversation?

“Alright, let’s get this over with. Rita, you start by stating your grievances.”

“I called you both here today to sort out your issues. Oliver, you can begin by explaining why you think your approach is better.”

“I appreciate both of you being here today. Let’s take a moment to set some ground rules and create a safe space for open communication.”

From option 3, I show this response (which I wrote, rather than having ChatGPT draft). ChatGPT did write a nice set of ground rules when I asked, which I could provide as a resource within the scenario.

You share a set of ground rules with Rita and Oliver. All three of you have a copy of the rules, and Rita and Oliver both agree to abide by them. You can view these ground rules any time in this scenario. (In the actual scenario, link to the ground rules resource.)

You ask, “Rita, would you like to share your perspective first, or shall we start with Oliver?”

To be continued…

This is a decent start for a draft scenario, at least enough to work with for a demo. I’ll continue showing my work for using AI to write the scenario in my next post.

Your experience?

Have you tried using AI to write scenarios? If you have any tips about what works or doesn’t work, let me know. We can experiment together.

If you’re interested in trying this yourself, check out these prompts for automating mini-scenario creation. That post has some good tips, although the scenarios generated all involve two people talking about a topic rather than making decisions and taking actions. The mini-scenarios are better than generic multiple choice questions, but they don’t provide decision-making practice.

Share this:

• Twitter
• Facebook
• LinkedIn
• Email
• More

• Reddit
• Pinterest
• Pocket

Related Tags: AI, Bard, branching scenario, ChatGPT, scenario-based learning, tools, writing Continue Reading Previous Comparing Tools for Building Branching Scenarios