Artificial intelligence’s (AI) aptitude to write computer code is supposed to help software developers navigate complex coding tasks and code more, but a Futurum report speaks contrarily.
Drawing on a 2024 DORA report by Google Cloud that highlights the impacts AI has on software development, the Furutum report titled “Can Developers Keep Pace With AI Innovations” states that there have been productivity decay among developers because of rampant use of AI.
According to the DORA survey it draws upon, for a majority (67%) of the 39,000 professionals it polled, AI has elevated their coding game, but for others, it has eroded productivity leading to instability in delivery.
A Stack Overflow study confirms this: Developers’ favorability towards AI development tools has fallen 5% since 2023. Now standing at 72%, it appears that as more developers have begun using AI tools in their everyday workflows, they feel more disappointed with the tools’ capabilities, or the lack thereof.
The area where most respondents strongly agreed to struggle is with finding knowledge with over half (53%) saying that waiting on information they need to get the job done is most disruptive to their workflow.
Despite the use of AI, knowledge siloes have continued to exist, taking away from developer experience and velocity.
Reports of GitHub Copilot introducing 41% more bugs into code shines additional light on the other possible reasons behind this sentiment.
The most common development tasks, the Futurum research finds, that companies rely on AI for are code writing, code reviewing and optimization, code explaining, codebase modernization, information summarization, documentation, debugging and data analysis.
“The software development technology market is experiencing a continuous stream of announcements of AI models, development tools using AI, AI Agents, and new AI-based automation capabilities,” wrote Mitch Ashley, vice president and practice lead of DevOps and Application Development at The Futurum Group, in the report.
“The pace of vendor announcements will not only continue; it will increase in volume, bringing increasingly disruptive benefits and changes to how software is developed, tested, secured, and deployed,” he added.
By next year, use is predicted to exceed 70% of organizations, according to the report.
“There will probably be too much investment, too quickly in areas of the economy, but if we fast-forwarded 30 years from today, the impact of this next generation of AI probably will meet the hype over time,” said Bret Taylor, cofounder and CEO of Sierra, an AI startup, in a Bloomberg interview about the trajectory of the AI wave.
But outcomes so far have been mixed. While AI tools demonstrate astonishing capabilities in terms of critiquing and improving code, studies after studies have shown that hasty deployment has cost dev teams time and speed.
39% of the respondents in the DORA study said they have low to no trust in AI. This is bad news considering that investment in and leaderships’ expectations from AI have only skyrocketed since the arrival of ChatGPT.
However, a vital cause of the mismatch seems to be leaderships’ lack of understanding of what developers truly require to be productive and what tools would make their job easier.
As companies rush to embrace more tools underpinned by GenAI to make coding faster and developers more productive, this finding reveals a major obstacle.
Based on another study by Atlassian on developer experience (DX), nearly half (44%) said they think their managers do not understand, nor are aware of the challenges they face in their jobs.
A whopping 69% said that over the course of a work week, they lose an average of 8 hours, if not more, to inefficiency – a problem they believe is completely avoidable.
“Developers, testers, and others are experimenting and using new AI offerings, which could contribute to defocusing on the work at hand. It’s a risk every manager would understandably be concerned with,” Ashley wrote.
Challenges of AI upskilling and ROI blind spots are other early stumbles leading to scant gains from AI adoption.
However, despite the pessimistic findings, AI is the best time-saving helper we have at hand. If utilized well, AI tools can bubble up problems hidden within code faster than any humans can. They offer easy suggestions to fix gnarly problems, autocomplete lines of code for developers, and even point to ways to improve them. The best tools can be the oracle for inexperienced developers and non-technical personas working across the organization.
But to reap those benefits in real-world, companies need to go beyond dropping AI tools willy nilly on developers’ lap without priming them or getting their approvals first. Thankfully, there are proven ways to ensure that developers’ productivity does not fall to the wayside from the frictions of AI adoption. The Futurum Report highlights five AI strategies that can help blend the technology into the fabric of the organization rather than upending it with AI. Such a successful transition can lead to many hours of freed time on developers’ schedules, and incremental gains in their productivity.
Read the full report here.
