Adoption of Large Language Models in Scrum Management: Insights from Brazilian Practitioners
This study presents an empirical analysis of 70 Brazilian Scrum practitioners, revealing that while Large Language Models are frequently and proficiently used to boost productivity in Scrum management, their adoption is tempered by significant concerns regarding output accuracy, confidentiality, and hallucinations.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are running a busy kitchen (a Scrum team) where chefs, servers, and managers work together to serve delicious meals (software projects) quickly. The paper you're reading is like a report card from 70 of these kitchen workers in Brazil, asking them: "How are you using these new, super-smart AI assistants (Large Language Models or LLMs) to help run your kitchen?"
Here is the story of what they found, broken down into simple parts:
1. The Setup: Who is using the AI?
The researchers asked 70 Brazilian software professionals. About half of them (49 people) actually use the Scrum method (the kitchen's way of organizing work). Out of those, 33 people said they are already using AI assistants like ChatGPT or Gemini to help with their management tasks.
These aren't just tech geeks playing around; they are serious users.
- Experience: Most of them are very comfortable with the AI. 85% say they are "intermediate" or "advanced" users.
- Habit: It's part of their daily routine. Over half (52%) use it every single day, often for an hour or two.
- Tools: Almost everyone uses ChatGPT, followed closely by Gemini and Copilot.
2. How are they using it? (The "What")
The paper found that the AI is being used like a super-powered sous-chef or a smart notepad, but not as the Head Chef. The AI helps with specific tasks, but humans still make the big decisions.
- Learning the Menu (Exploring Scrum): This is the most popular use. People use the AI to ask, "How does this Scrum rule work?" or "Can you explain this concept?" It's like having a encyclopedia that talks back.
- Writing the Orders (Artifacts): The AI is great at writing things down. People use it to draft "Product Backlog Items" (the list of things to build) or to fix up "Sprint Goals." It's like the AI is a fast typist that helps organize the to-do list.
- Summarizing the Meeting (Events): After a team meeting, the AI helps summarize what was said. It's like a secretary who takes notes and highlights the important parts.
- What they don't do yet: The AI is rarely used for the "soul" of the team. People are hesitant to let the AI decide the "Product Vision" (the big dream of the project) or to fix "Working Agreements" (how the team treats each other). These require human empathy and judgment, which the AI isn't trusted with yet.
3. The Good Stuff (Benefits)
The workers reported that the AI is a huge time-saver.
- Speed: 78% said it makes them more productive.
- Less Tedium: 75% said it reduces the boring, repetitive work.
- Better Notes: 78% said the quality of their written documents (like requirements) improved.
- Brain Relief: It helps them think through problems without getting stuck, acting like a "thinking partner."
4. The Bad Stuff (Risks & Warnings)
Even though the AI is helpful, the workers are very cautious. They see it as a tool that needs a human supervisor.
- The "Almost Right" Trap: The biggest complaint (81%) is that the AI gives answers that look correct but have small, dangerous errors. It's like a sous-chef who chops vegetables perfectly but forgets to wash them. You have to check everything.
- Hallucinations: 59% said the AI sometimes makes things up completely (hallucinations).
- Secrets: 63% are worried about privacy. They don't want to accidentally paste secret company data into a public AI chat.
- Trust Issues: Because the quality varies so much, you can't just let the AI run the show. You have to watch it constantly.
5. The Big Picture Conclusion
The paper concludes that Brazilian Scrum teams have adopted AI like a reliable but quirky assistant.
- They love it for: Writing, organizing, learning, and summarizing.
- They fear it for: Making strategic decisions, handling secrets, or being 100% accurate.
The Golden Rule: The AI is a great tool to speed up the work, but the human team must remain the "Head Chef." They need to check the AI's work, keep secrets safe, and make sure the final decisions are still made by people, not machines.
In short: The AI is a powerful new kitchen gadget that helps you cook faster, but you still need to taste the food before you serve it to the customer.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.