Experimenting with AI-assisted first steps for journalism students

My colleague Liz Hannaford came back from this year's Association of Journalism Educators (AJE) conference, with loads of interesting ideas, especially around experiential learning and new technology like AR/VR and, inevitably AI.

One presentation in particular caught our attention. Daisy Naylor and Julia Breen from Sunderland Uni talked about using generative AI to create a choose-your-own-adventure-style game covering disaster reporting.

It was a really interesting idea and one that prompted some thought and experimentation. Liz took the idea and created a prompt based on the scenario of covering a local swimming pool closure. Liz is primarily focused on the audio storytelling side of things, hence the lean into that in the example prompt. Thanks, Liz!

The scenario is not as exciting (or inaccessible) as the disaster story. But no less nerve-wracking an experience for students starting on reporting. For overlapping reasons, they are safe places to explore. It's also a good bread-and-butter story to explore in terms of process. It's also a great target for the type of story a student could (and should) be comfortably turning around by the end of their first year (or sooner)

With some playing around, we had a prompt that seemed to work (and was fun to explore). You can access the full prompt via Git Hub.

This is the kind of thing it generates (using ChatGPT 3.5). This is a snippet from the story as it develops.

You prepare interview questions focusing on Sarah's personal experiences with the pool and how its closure would impact her family. You jot down notes and head out to meet her at a local cafe where she suggested to meet.

Sarah arrives, visibly concerned, and begins sharing her story about how the pool closure will affect her children's swimming lessons, which they rely on for physical activity and social interaction.

{What do you do next?}
Option 1. Record Sarah's interview using your smartphone, capturing her emotions and key points.
Option 2. Ask Sarah if she knows others who are similarly affected and could provide additional perspectives.
Option 3. Take a photo of Sarah to include in your multimedia report, with her permission.
Option 4. Something else

The prompt is set up to allow you to type file story and then it will outline the choices you made, offer some feedback and suggest what a story might look like.

Summary of Decisions Made:
1.Visited social media platforms to gauge public sentiment.
2.Contacted and interviewed Sarah, a local resident affected by the pool closure.
3.Recorded Sarah's interview to capture personal perspectives.

News Story:
Local Community Divided Over Future of Mediatown Swimming Pool

In the heart of Mediatown, the impending closure of the Mediatown Swimming Pool has sparked a wave of emotional reactions from residents. Sarah Johnson, a local mother of two, expressed her dismay over the decision, emphasizing how the closure would impact her children's access to swimming lessons and community recreation.

"The pool isn't just a place to swim; it's where my kids learn water safety and socialize with friends," Sarah lamented.

In contrast, others in the community argue that the closure is a necessary measure due to budget constraints and the need to allocate resources to other municipal services.

The Mediatown Council has yet to provide a definitive statement on the closure, leaving residents uncertain about the pool's future.

Editor's Feedback:

This story effectively captures the personal impact of the pool closure through Sarah's interview and provides a glimpse into the broader community sentiment. The inclusion of social media reactions and a balanced perspective from residents adds depth to the report. To enhance future reports, consider:

Diversity of Sources: While Sarah provided a compelling personal account, diversifying sources could offer broader perspectives, such as from council members or local businesses affected by the closure.

Multimedia Integration: Including more multimedia elements like video clips or interactive graphics could further engage our audience and enrich the storytelling.

Follow-Up Potential: Explore opportunities for a follow-up story, such as investigating the financial implications of keeping the pool open versus closing it.

Keep up the excellent work, Alex! Your dedication to capturing local stories is invaluable to our readers.

ChatGPT was very flexible in the interaction. I tried entering two options. For example, from this initial prompt.

{You are Alex Johnson, a final-year multimedia journalism student interning at the Mediatown Times. Your mission is to create a compelling multimedia news story about a threatened pool closure. Each decision you make will shape your path. When you think you have finished reporting, you can type “File Story” to send your work to the editor. Choose wisely, and may the scoop be with you!}

{What do you do?}

Option 1. Visit the local swimming pool to interview regular visitors and staff.

Option 2. Attend a town council meeting to gather official statements and perspectives.

Option 3. Talk to local business owners about the impact of the pool closure on the community.

Option 4. Something else.

Two possible options look good - I could do both!

I inputted 1 and 3 and it gave me this...

You decide to start by visiting the local swimming pool to interview regular visitors and staff, as well as talking to local business owners about the impact of the pool closure on the community.

Visiting the Swimming Pool
You arrive at the Mediatown Swimming Pool, a bustling hub for families and individuals seeking exercise and community interaction. The atmosphere is lively, with children splashing in the shallow end and adults swimming laps in the deeper sections. You approach a group of regular visitors to begin your interviews.

Interviewing Regular Visitors and Staff
You meet Emily, a long-time visitor who brings her children to swim every weekend. She expresses her concern about losing such a valuable community resource.

"I can't imagine where else we would go for affordable family fun and exercise," Emily says, her voice tinged with worry.

Next, you speak with Mark, a lifeguard at the pool, who shares insights into the safety measures and the pool's role in teaching swimming lessons to children from local schools.

"The pool isn't just about recreation; it's about teaching kids a life skill that could save lives," Mark emphasizes.

Talking to Local Business Owners
After gathering perspectives from the pool, you head to nearby businesses to understand the potential economic impact of the pool closure. You speak with Mr. Patel, who runs a cafe across the street from the pool.

"Our business thrives during swim meets and weekend visits. If the pool shuts down, it will definitely hurt our sales," Mr. Patel explains, visibly concerned.

You also chat with Mrs. Carter, who manages a sports equipment store nearby.

"The pool closure would mean fewer families coming into the area. It's not just about the immediate loss of customers; it's about the community vibe changing," Mrs. Carter remarks thoughtfully.

What do you do next?
Option 1. Attend a town council meeting to gather official statements and perspectives.

Option 2. Check social media platforms for reactions and comments from local residents about the pool closure.

Option 3. Compile your notes and prepare to return to the Mediatown Times office to start drafting your multimedia report.

Option 4. Something else.

ChatGPT is good at handling mixed answers

A pretty solid round-up! And ChatGPT is smart enough to also take text entry too. So you can go off-piste and put in your responses long hand.

How and why would this work?

Most of what we do in journalism schools is learning by doing. In short, there's no replacement for going out and covering stories - especially ones like the swimming pool story. I know people from industry, in particular, might be a little suspicious of this approach in that context. But, using AI in this way is all about context and the learning journey.

Imagine this is an exercise with first-year journalists fresh in the room. Getting them to work it through in groups, comparing each group's responses and the different outputs and advice is a great experiential learning exercise. They can experiment and hopefully remember the conversation. Add an example of a real package in a similar vein and you're driving some proper reflection.

Extending the idea - virtual job interviews.

The exercise also got me thinking about other similar activities that could benefit from this approach. One I tried was using it to create a prompt for a short job interview scenario.

I found a job description for a multimedia reporter job online and used the text as part of a prompt for a short screening interview.

The following is a job description for a reporter role at a UK media organization that focuses on audio output.  

<TEXT OF THE JOB DESCRIPTION>
 
Using this as source material, create a short, interactive text based, choose your own adventure style job interview where you play the news editor and I play the candidate. It should consist of five questions, presented with four possible answers. After each question, wait for my choice before proceeding. Do not repeat the questions. When you have asked all five questions, you should recap the answers with feedback on the choice and offer general feedback on my performance at the interview.  

The prompt for an AI-driven job interview.

Here's an example of the output from ChatGPT3.5

Welcome to the interview for the Multimedia News & Sports Reporter role. I'm excited to learn more about your skills and experiences. Let's begin!

Question 1: How do you prioritize your tasks when reporting on multiple platforms simultaneously?

a. I create a detailed schedule to ensure each platform receives timely updates.
b. I focus first on on-air reporting, then adapt the content for online and social media.
c. I prioritize breaking news, adjusting my approach based on audience engagement.
d. I collaborate closely with the team to delegate tasks effectively.

Please select the answer that best reflects your approach.

An example of the output from ChatGPT3.5

It's not as rich or immersive as the previous example. But again, as a tool to prompt thinking about what areas to focus on, I think it's a good experiential exercise. If nothing else, it might prompt some thought on possible responses.

Choosing the right platform.

There are lots of issues here. The biggest is the inconsistent way the platform interprets the prompt. Each prompt elicits a slightly different interaction. This is bad enough on one platform. The real challenge is how these things might be disrupted when you are using more than one platform.

When discussing AI tools with students, I tend to focus on three core generative AI platforms - ChatGPT, Gemini and Co-pilot. These are accessible (free for the most part) and offer a chance to compare and contrast responses. It may also be that institutions begin to lean towards one or another. At MMU for example, the policy on Generative AI suggests that CoPilot is the platform we recommend to students as they all have access to it via their student office accounts. It makes for a level playing field. But you may be in a Google shop where Gemini access is the norm.

However, I thought it was worth playing with the prompts and looking at the quality of the responses. Here's my summary:

  • ChatGPT (version 3.5) both the scenario and job interview prompts worked well. There were times when the responses strayed. But for the most part, the experience was consistent.
  • Gemini - The responses to the scenario prompt were a bit hit-and-miss not least because the connectivity to the web meant it tried to add live context to the content. More alarmingly it had a very lax attitude towards interviewing minors and swanning around a pool taking candid pictures! It faired much better in the job scenario where the interaction felt better and more constructive than ChatGPT. It also threw in questions about OFCOM and regulation even though that wasn't mentioned in the job ad!
The Manchester Aquatics Centre sounds like a good place to start - it was used in the 2002 Commonwealth Games and has a public pool.  You head there, weaving through the bustling crowds.

The chlorine scent hits you first, a wave of nostalgia washing over you.  The air is loud with the echoing shrieks of children and the rhythmic thwack of pool noodles against the water.  Parents shepherd their little ones poolside, their faces etched with worry lines.  A group of teenagers huddle by the diving boards, their laughter tinged with defiance.

You approach a young family with a little girl clutching a pink inflatable flamingo.

Option 1.  Introduce yourself and explain the situation. Ask the parents how the closure of the pool would affect their family.
Option 2.  Strike up a conversation with the teenagers. See if they're part of a swimming club and what impact closure would have on their training.
Option 3.  Mingle with the crowd and try to capture the general mood through candid photos and ambient sound recordings.
Option 4. Something Else

What do you do?

Eeek! Gemini (and Co-pilot) don't do the ethical/legal stuff well at all!

  • Co-pilot. When it worked it was comparable to ChatGPT (as I guess it should be). But it was more like its easily distractable younger sibling! Like Gemini, it got thrown off track by trying to include web searches and other online material. It also struggled with the format - In both cases, repeating questions and making the possible responses so obvious that there was only one real answer. In one iteration of the interview questions, it abruptly said it was stopping. I'm guessing because the content it had generated was becoming contentious.

Conclusions

A lot of the issues/challenges/play are in the creation of the prompts. One prompt is not going to fit all platforms. And there will be loads of suggestions for ways to improve ours. In a classroom situation, you'd need to be prepared to smooth over the weird responses. But battling for consistency learning across the many different platforms we use is something anyone who has worked in journalism and journalism education will be depressingly ready for!

Generally, though I think some experimentation and some properly guided/scaffolded exercises would make this kind of use a great way to explore some of the basics, especially for colleagues in j-schools who lean on the experiential learning pedagogy style of things. For those in the industry (or academics with a more situated view) don't be put off. As a prompt for practical reflection, it did make me think about conversations I've had around story development both with students and in real life.

The prompts are available in full

If you're doing or have already done something similar (sorry!) let me know, I'd love to see more examples.

Massive thanks Liz for reporting back on an interesting idea, having a go and sharing. And a massive hat tip to Daisy and Julia for the creative thinking and ideas.