The Impact of AI on the Scrum Team

Generative AI is rapidly changing many industries, and the Agile world is no different. Tools like OpenAI’s ChatGPT are evolving and starting to impact the roles within Scrum teams. This article looks at how AI is transforming the jobs of ScrumMasters and Product Owners and what this means for the entire Scrum process. By comparing these changes to those in the pharmaceutical industry, we can better understand what the future might hold for Agile methods influenced by AI.

The Scrum Master

If you’re like me, you probably started using generative AI in stages. At first, it seemed like the technology advanced too quickly. Then, you tried OpenAI’s ChatGPT, maybe asking it to write a post or summarize an article, and were amazed at how quickly it produced decent content. I did the same, asking questions about Agile coaching and summarizing articles. That’s when it hit me: this technology could potentially replace my job.

As ChatGPT evolved from version 3 to 3.5 and then 4, each new version reinforced my belief that some jobs, including mine, might become obsolete. It might not happen today or tomorrow, but it’s likely in the future.

Navigating a Fast-Moving Agile Environment

If you’re part of an Agile team, you’re working with a fast-moving, ever-changing product. To succeed, you need to adapt to these changes and adjust your career accordingly. It’s similar to a line from “The Godfather Part II” where Hyman Roth tells Michael Corleone, “This is the business we’ve chosen.” If you want to be involved in Agile work, you need to be prepared for significant changes. This mindset is essential for all Agile team roles, but it’s especially crucial for the ScrumMaster role.

The Vulnerability of the ScrumMaster Role

Many might be surprised that the ScrumMaster role is vulnerable because it’s a celebrated position. Millions have pursued ScrumMaster certification, and the Scrum Guide describes this role in detail. Yet, the Scrum Guide defines the ScrumMaster role as involving administration, coaching, and training — skills that AI-enhanced Agile tools are starting to integrate.

The Future of the ScrumMaster

While the ScrumMaster role will likely persist for a long time due to its reliance on soft skills like communication, problem-solving, critical thinking, and fostering creativity, its necessity might diminish in smaller Agile teams. Traditionally, the ScrumMaster role combines servant leadership with administrative tasks, such as coordinating team meetings. As administrative tasks reduce, a significant part of the role may disappear.

Organizations might start having one ScrumMaster support multiple teams, similar to Scrum scaling frameworks like Large-Scale Scrum but in a scaled-back manner. If fewer ScrumMasters are needed, the demand for this role will decrease. Although ScrumMasters won’t vanish, the demand will lessen, leading many former ScrumMasters to pursue other opportunities.

To stay relevant, it’s crucial to enhance your skill set before these changes take full effect. As you’ll see later, new and exciting opportunities will emerge for those who wish to remain part of an Agile team.

The Product Owner

In 2023, the company behind ChatGPT conducted a study to assess the impact of their models on the job market. They tested a large language model (LLM) on about 19,000 tasks across 900 different jobs. The results showed that the model could perform approximately 80% of the tasks. Interestingly, the more education a job required, the more likely it was that an LLM could do the work. For jobs requiring a four-year degree, the model could handle 74% of the tasks, while for jobs needing a graduate school degree, it managed 64%. This indicates that generative AI is more likely to replace experienced, well-educated employees than lower-paid or entry-level workers.

That same year, the United Kingdom’s Department of Education conducted a similar study. They ranked jobs based on their likelihood of being impacted by generative AI. Management consultants and business analysts topped the list. If you’re a product owner, this might catch your attention, as product owners and business analysts share many similar tasks and skills.

Product owners, like business analysts, interact with customers and determine what should be included in the product. Their tasks often involve communication, meetings, product development, and summarizing information. Given these similarities, it’s clear that generative AI could significantly impact product owners as well.

The Evolving Role of the Product Owner

Currently, product owners do much more than just creating user stories. However, in the near future, organizations will need to decide how much more these roles will entail. Similar to the ScrumMaster role, product owners won’t disappear overnight, but long-term opportunities may decrease. If an LLM can generate most user stories, the product owner’s role might be reduced to editing and prioritizing them.

AI-Enhanced Workflow

Imagine an AI software repository that writes most of the user stories based on information extracted from meetings and emails. The AI system could also recommend priorities and even insert chunks of working code. AI-enhanced coding tools will likely boost developer productivity, meaning teams might only need one or two developers. In this scenario, organizations might not hire a full-time product owner for each small Scrum team. Instead, one product owner might manage several teams, similar to existing Scrum scaling frameworks.

The main difference is not that Scrum is expanding too much, but that many of the product owner’s tasks will be automated. Intelligent repositories and LLMs in software development will likely reduce the long-term demand for product owners. Much like business analysts, many of their tasks will be automated by newer generative AI systems. Consequently, there will be fewer product owners, leading to less administration and a greater focus on data, vision, and creativity.

This shift means there will be excellent opportunities for former product owners who wish to transition to roles in research and data. As you’ll see later, these changes will alter the overall makeup of Scrum teams.

Virtual Sprinting

You’ve seen how generative AI affects the roles and responsibilities of your Scrum team, but it’s also important to consider its influence on the entire Scrum process. To understand this, let’s look at changes in the pharmaceutical industry, which also relies heavily on experimentation.

Developing a drug involves identifying promising compounds and testing them, a process that traditionally requires many people and extensive real-world experiments. However, cutting-edge AI systems can now simulate these experiments. Instead of using microscopes, beakers, and test tubes, hundreds of compounds undergo rapid virtual experimentation within a neural network. Once a promising compound is found, the AI makes small changes and conducts more experiments swiftly. This advancement has reduced the drug development timeline from years to months.

This pharmaceutical process is similar to how Scrum teams develop software. Imagine your team using Microsoft Copilot, an AI system that can rapidly create working software. Currently, experienced human developers input code like print “Hello World!” and the system generates more code. As technology improves, it will generate larger blocks of code based on simpler prompts, such as “Create code to efficiently access our database from a smartphone.” This could lead to software generation with little or no manual coding.

Virtual Development Environment

So, what does this mean for the Scrum experimentation process? It could become similar to the pharmaceutical industry’s virtual experimentation. Instead of human developers assembling blocks of code, everything would happen in a virtual development environment. The AI system could conduct many virtual sprints in a simulated setting, generating the best version of the product.

You can think of this like Doctor Strange from Marvel Comics, who travels through the multiverse and takes the correct path because he’s seen the possibilities of each step. A virtual coding environment could try different blocks of code and improve the product’s efficiency step by step. An AI system could go through thousands of product versions until it finds the most efficient one, a process that has traditionally been painstaking for human developers.

What does rapid experimentation and delivery mean for your Scrum team? Fewer developers will be able to conduct multiple experiments, and the software will likely be more complete when it emerges from the AI environment. In the future, Scrum teams with just one or two people could produce much more software. Scrum processes that support feedback and experimentation could be handled in a virtual environment driven by an AI system.

Conclusion

In summary, AI is set to significantly change the roles within Scrum teams and how the Scrum process works. As AI continues to advance, many tasks currently done by ScrumMasters and Product Owners will likely be automated, which could reduce the need for these roles. However, this also creates new opportunities for those who are willing to adapt and learn new skills. The future of Agile will feature smaller, more efficient teams that can quickly experiment and develop products in AI-driven virtual environments. Adapting to these changes will be essential for staying relevant and successful in the evolving world of Agile work.

For a deeper understanding of the foundational concepts, you might find it helpful to read AI Enhanced Agile Practices: Artificial Intelligence In Action