Asking developer teams if AI is helping them be more productive usually results in a response of "meh, it's OK."
What?!? AI represents the biggest technological advancement since the internet, and it's meh?
This attitude is backed by research in Atlassian's State of Developer Experience report, where two-thirds of engineers report they have yet to experience productivity gains. This was sharply contrasted by their leaders, who believe AI to be the most effective way to improve productivity and satisfaction.
It turns out that these two statistics are more connected than we first thought.
Before we explore that, let's take a look at why developers might not be seeing as many productivity gains as they could be.
When most people think about using AI for software development, they immediately think of AI code generation or code completion. For this reason, it's understandable that seasoned developers aren't as likely to realize huge productivity gains when it comes to AI-assisted coding.
This is because they are already expert coders, and coding isn't a point of friction for them. It does help to remove some repetitive tasks, though, so it provides some value, just not game-changing. In other words, meh.
However, new research shows us that there's more to the story.
It turns out that how developers use AI, not if they use it, makes a big difference in the meh factor. Research indicates that strategic AI collaborators -- those who work with AI as a teammate, not a tool -- report more time savings and better work quality.
There are also simple AI users, who treat AI just as a personal assistant that can automate basic tasks and save them some time. In this stage, folks are using AI for things like code completion or optimization. Most AI users are currently at this stage, where they're experiencing some gains but the potential is mostly untapped. During this 'meh' stage, folks are saving around 53 minutes per day.
Progressing from a simple AI user to a strategic AI collaborator is where the magic happens.
Strategic AI collaborators see AI as a team member to spar creatively with and potentially learn from. This requires a change in mindset from AI being an automation tool to AI being a teammate. If we stick with the coding example, it means going from having AI generate code for you to correct, to partnering with AI like you would with another engineer in paired programming.
Asking why it gave you certain feedback and the pros and cons of different approaches helps an engineer get to a higher quality outcome faster. When it is time to work with a human teammate, engineers work through fewer review cycles because the output from the solo engineer is higher quality.
Strategic AI collaborators report saving on average 105 minutes per day, and a 90% improvement in the quality of their work -- nothing meh about that!
Engineering leaders say AI is the most effective way to improve developer satisfaction and productivity. If they want their teams to realize those productivity gains, they'll have to show them the way.
Reflecting on my own experience, my leader often talks about the different ways he uses AI and encourages me to do the same. This has changed my behavior in two different ways; I feel more comfortable using AI so I use it more, and I also encourage my team to use AI. This approach is backed by Atlassian research which shows leaders who encourage team members to experiment with AI save 55% more time, and people who have leadership support to use AI are more likely to be strategic collaborators.
This is a theme throughout Atlassian. We're actively working to create a positive shift to having a large group of strategic AI collaborators. Senior engineering leaders often share about their team's experimentation with AI and encourage other engineers to do the same. This has led to greater experimentation with AI across engineering, with the results getting embedded in our products for customers to use.
Encouraging experimentation with AI has resulted in some fantastic use cases for AI within Atlassian, that customers are also benefiting from. Pull Request (PR) cycle time has been a challenge for our engineers at Atlassian, so our teams experimented with ways they could use AI to speed things up. They created AutoReview, an AI-powered code review assistant that reviews code changes, identifies potential issues, and suggests improvements. AutoReview facilitates engineer AI collaboration on PRs, improving the efficiency and quality of code reviews whilst improving developer satisfaction and productivity.
It's clear that engineering leaders have a huge role to play if they want their teams to realize the potential productivity and quality gains that come from using AI.
Considering AI as a virtual teammate, experimenting, and sharing your results with your teams will help set them on a fast path out of the meh zone and toward becoming strategic AI collaborators.
