When ChatGPT Writes the Stories, Who’s Thinking About the User?

We’ve trusted Jira to manage our user stories for years and now it seems ChatGPT has started writing User Stories for us! That sounds great, doesn't it? Yeah! Maybe in a world where ChatGPT is the user too. In that world, ChatGPT knows the mind of ChatGPT. Wait a minute, ChatGPT knows its mind? It has a mind? well, I digress - that’s a different topic altogether for another time. (watch out for another article)

It seems that some product development teams are already turning to ChatGPT to write user stories for them. There are many Youtube videos instructing you how to write user stories with your favorate LLM. There are even tools available today given an idea they will spit out stories! Why stop there, once a story is written, LLM can write decent test cases and more. Imagine a future where the LLM not only writes the user stories but also codes the feature, tests it, and packages it into a product. It can certainly create marketing campaigns, possibly sell the new product online, and all that’s left for you to do is sit on a tropical island and collect the profits. It sounds like a dream, right? ?

It really not that far-fetched. Today’s LLMs are already helping with coding, test cases, marketing copy—you name it. With a bit of orchestration (agents?), we could automate even more in the near future. This might lead to a scenario where we could churn out product after product, all powered by artificial intelligence.

Hmm... have we seen this before? Maybe not at this scale, but we’ve definitely seen some ambitious tech-driven efforts in the past. And they’ve taught us a thing or two... or did they?

The Danger of Forgetting the User

?In the rush to leverage LLMs and automation, we might forget to ask the most critical question about our products: Is this something users actually want? History shows that even with great technology and automation, missing the mark on user needs can lead to failure.?

Take the example of Google Glass, a product that had cutting-edge technology but failed to resonate with users. It wasn’t clear how the device would fit into people’s everyday lives. Then there's Microsoft’s infamous virtual assistant. Ah, Clippy... remember him? Maybe too helpful for his own good. Despite its intelligent design, Clippy’s attempt to predict user needs was often frustrating, and users never really adopted it.

Start with the User

So, where does that leave us? The truth is that while LLMs can give us a leg up—particularly in generating ideas and creating variations of concepts—they can only go so far. These models are great at combining existing ideas or suggesting novel combinations. They can also patiently sift through defects or user comments to suggest product improvements, even detecting user sentiment from feedback. But do they have the ability to imagine user needs? In reality, humans struggle with that too—that’s why we engage the actual users in the process.

In product development, LLMs can be powerful when directed by the user, but it's essential to keep the human as the driver. For instance, they can help analyze and organize large sets of feedback or generate early-stage ideas. Still, the best move any team can make is to validate those ideas with real users as soon as possible.

In fact, this is something I cover in my workshops —come to learn more about how to effectively integrate AI into product development without losing sight of the user. We will touch on that at the upcoming meetup event Unveil the Mysteries of ChatGPT: Revolutionize Your Product Development with AI hosted by Angela Johnson . The key is using LLMs as a tool to amplify human creativity and decision-making, not replace it. At least that's where it stand today.

As Ron Jeffries —one of the three founders?of Extreme Programming— articulated that, while user stories might be written down, the writing [of user stories] is not the most important part. In fact, less important than the thinking, communicating, and testing that goes into truly understanding what needs to be done. This underscores the importance of staying focused on the users, involving them in the process, and continuously validating their needs.

So, where do you stand? Are we prepared to rely on an LLM to create stories for you? I’d love to hear your thoughts—have you used LLMs in your product development process, and what’s your experience been like?

Reference:

1. How should user stories be written ? by Ron Jeffries

More Actions to learn

Join a community to explore these ideas more

Agile Denver Meetup Group ’s? SIG for AI

The AI Product Development Network - a Meetup group

Attend a Workshop

Product Development with AI

Note

This article was authored by Manoj Vadakkan , with assistance from ChatGPT4. All original ideas and the initial draft were created by Manoj.