Google: AI and Platform Engineering Are Transforming DevOps

More than 75% of working professionals worldwide use AI at least once daily for work, but far fewer trust AI-generated code, according to a survey of 3,000 employees in Google’s 2024 Accelerate State of DevOps Report (DORA).

The study, published on Oct. 22, revealed that 76% of professionals use AI to write code, summarize information, explain unfamiliar code, optimize code, and document code. It outlined the many benefits of generative AI adoption, including increased focus, productivity, job satisfaction, and code quality.

However, generative AI can also negatively impact software delivery performance, product quality, and the time employees spend on valuable work, the report indicated. It also found that using AI does not necessarily reduce time spent on “toilsome work,” or tasks that lack “meaningfulness.”

“AI has positive impacts on many important individual and organizational factors which foster the conditions for high software delivery performance,” the report states. “But, AI does not appear to be a panacea.”

Google survey outlines pros and cons of generative AI

This year’s study, the 10th iteration, focused on how AI impacts burnout, focus, job satisfaction, productivity, and the performance of products, organizations, and teams. DORA measures stability success through four key metrics: change lead time, deployment frequency, change fail rate, and failed deployment recovery time.

SEE: Choose from our list of 10 best project portfolio management (PPM) software and Tools for 2024.

Interactions with AI in daily work tended to come in the form of:

• Chatbots (78.2%).
• External web interfaces (73.9%).
• AI tools embedded within their integrated development environments (72.9%).

Some respondents reported adopting AI in response to competitive pressures, with one interviewee noting that companies not embracing AI risk being “left behind.” Another mentioned their organization viewed AI as “a big marketing point.” Fewer than 10% of respondents said their productivity had been negatively impacted by AI.

Additional findings show:

• 81% of respondents said “their organizations have shifted their priorities to increase their incorporation of AI into their applications.”
• Developers feel more productive when using AI, with 67% of respondents reporting that AI helps them improve their code.
• Nearly 40% of respondents said they had “little to no” trust in AI.

On the other hand, a majority of respondents said they only “somewhat” trust the quality of AI-generated code. Interviews, as well as the study’s authors’, indicate this may mean developers expect to use AI as a baseline from which to tweak and correct the results.

“However, respondents also reported expectations that AI will have net-negative impacts on their careers, the environment, and society, as a whole,” the report reads. Over 30% of respondents think AI will be detrimental to the environment.

AI may also impact software delivery performance, stability, and throughput. This may be because AI-written code can be generated in such large amounts. These larger changes are “slower and more prone to instability,” according to the report. Small batch sizes are still an important principle in software development that directly relates to quality.

Nearly 9 in 10 professionals use internal developer platforms

Platform engineering is a discipline for creating workflows to promote self-service and collaboration. DORA describes it as the intersection of social interactions between teams and technical performance — such as automation, self-service, and repeatability of processes.

DORA found that 89% of respondents used internal developer platforms, although the definition of the term was left quite broad. The report also found:

• Organizations tend to see performance gains at the beginning of a platform engineering initiative, followed by a dip and a leveling out. This pattern matches that of other transformation initiatives DORA studies.
• Individuals were 8% more productive when using an internal developer platform.
• Organizations performed 6% better when using an internal developer platform.
• Throughput and change stability fell by 8% and 14%, respectively, when using an internal developer platform.

Why such a large drop in change stability? DORA suggests the platforms could increase rework time. Or, this number could be indicative of a different pattern: teams with high pre-existing instances of burnout and change instability may adopt platforms to solve those problems.

Additional findings include the importance of stable priorities

The complete report goes into more detail on these topics. Additional takeaways include:

• Product quality is proportional to how well the organization understands its users’ needs. User-centered software development is beneficial because deriving a sense of purpose — directly meeting users’ needs — benefits both employees and organizations.
• Organizations should give developers confidence that their projects are meaningful — a process that requires user feedback.
• Focus on creating quality documentation. This is documentation that is not necessarily comprehensive but instead is relevant, findable, and reliable.
• Unstable priorities can cause burnout in its employees. Namely, “move-fast-and-constantly-pivot” mentalities from leadership can hurt employees. This mindset creates unclear expectations, decreases employees’ sense of control, and increases their workloads.
• Leaders should be positive. While they can still challenge their workers to think innovatively, leaders should also recognize employees’ successes.

“The key to success is rolling up your sleeves and just getting to work,” the report stated. “The goal of the organization and your teams should be to simply be a little better than you were yesterday.”