Top insights for IT pros
From cybersecurity and big data to cloud computing, IT Brew covers the latest trends shaping business tech in our 4x weekly newsletter, virtual events with industry experts, and digital guides.
The majority of professional software engineers will be using AI assistants by 2028, Gartner projects—but tech execs expecting massive increases in productivity are likely to be disappointed.
According to Gartner, while under 10% of software devs were using such assistants in early 2023, another survey in late 2023 showed 63% of organizations were piloting or deploying them. By 2028, the consulting firm expects three out of four devs to be using AI assistants regularly on the job.
“You see the jump,” Philip Walsh, a senior analyst in Gartner’s software engineering practice, told IT Brew.
The number of devs using AI assistants had already risen from under 10% in early 2023 to 18% by October, Walsh said, and an additional three-fifths of respondents reported their organization was planning, piloting, or deploying AI tools. That said, he cautioned productivity gains in practice might be modest, even if some of the most optimistic projections—like time savings of up to 50% on some tasks—are correct.
“It’s very difficult to turn time saved into time reapplied,” Walsh said. “With an AI code assistant, the time savings are coming in five, 10, 15, 20-minute increments…If time savings don’t come in a certain large enough threshold, your developers aren’t going to immediately just launch into the next ticket in the backlog.”
Though AI assistants do improve developer experience and help them stay focused, Walsh said, developers actually spend a relatively small portion of their time writing new code or improving existing code (32%, according to The New Stack). According to Walsh, AI assistants primarily assist developers with tasks like researching solutions and writing “more boilerplate” code that benefits from auto-completion.
“The CFO, the CEO, they just hear this number: ‘Developers are getting work done 50% faster, so that means they’ll be 50% more productive,’” Walsh told IT Brew. “But that’s a false inference, because it relies on the implicit premise that developers spend 100% of their time coding.”
According to Walsh, organizations that are already mature in terms of best practices like continuous integration/continuous deployment automation, DevOps automated testing, and agile development may be better able to extract productivity gains from generative AI. Yet Gartner polling of software engineering leaders, he said, has shown the biggest productivity gains come from another category of software entirely—“communication and collaboration tools.”
“Software development really is a team sport," Walsh said. “There are a lot of handoffs, there’s a lot of just sending things to people, having things reviewed, rework and revisions being done, so anything that improves that comms cycle actually has a far bigger impact on productivity than a tool that just supports, that’s more narrowly task-related.”