If you work with software, in any industry, then you’ll be using gen AI before you know it, if you aren’t already. This is the conclusion of an interesting study by Capgemini into the most hyped technology out there. And the early results are in: organisations with active generative AI (gen AI) projects are already seeing benefits in terms of faster innovation, better quality, higher productivity, and less time. The problem, of course, is these testimonies come from an extremely limited pool; hardly any enterprises are properly engaged with generative AI yet.
The Capgemini study, actually published a month ago, polled a thousand-odd “senior executives” and another thousand or so “software professionals”, including architects, developers, testers, and project managers. There is no detail on its vertical split, by industry sector. It found that generative AI is to be adopted by 85 percent of the “software workforce” over the next two years. But importantly, nine in 10 enterprises are “yet to scale” with the technology.
As such, the reported upside from adopting generative AI – 61 percent said it “fosters innovation” and 49 percent said it improves software quality, apparently; productivity improves by seven-to-18 percent, they said; time savings were “as high as 35 percent”, they said – is claimed by a small number of early adopters, which might presumably be imagined as large corporations with deep pockets. (There is no demographic data or criteria about the sample.)
But the survey is interesting insofar as it explains how the enterprise sector – whichever parts of it that were quizzed in the study – perceives the likely usage of generative AI. Eighty percent of software professionals think generative AI tools will “significantly transform their function” by automating simpler repetitive tasks. More than 25 percent of software design, development, and testing work will be done with generative AI in the next two years, they say.
This will free software engineers to focus on “higher-value-adding tasks”, says Capgemini. About half reckon this means time to develop new software features (50 percent) and generally upskill (47 percent). Interestingly, as well, reducing headcount appears to be the last thing on the minds of enterprises. Just four percent of “responding organisations” picked this as the strategic upside to deploying generative AI (“the least-adopted route”).
More than this, new roles for generative AI developers, prompt writers, and architects are emerging, says Capgemini. Also of interest; the perception among software engineers is that generative AI will “boost their comms with… non-technical business teams”, it says. More than three quarters (78 percent) are described as “confident” and “optimistic” that enhanced collaboration will be the result.
Despite the almost-total lack of planned and scaled enterprise-designed initiatives, generative AI tools are used today by 46 percent of software engineers to assist with tasks, mostly for code-writing. But almost three quarters say its potential extends beyond just writing code. “While coding assistance is the leading use case, generative AI also has applications in other… activities, such as code modernisation or user experience design.
Problem is, the survey finds that 63 percent of software professionals admit to using unauthorised generative AI tools. It states: “This rapid take-up, without proper governance and oversight in place, exposes organisations to functional, security, and legal risks like hallucinated code, code leakage, and IP issues.” Reason to invite Capgemini in, of course; Pierre-Yves Glever, head of global cloud and custom apps at the firm, explains the pitch.
“Its impact on coding efficiency and quality is measurable and proven, yet it holds promise for other software activities. However, we must remember that the true value will emerge from a holistic software engineering approach, beyond deploying a single ‘new’ tool. This involves addressing business needs with robust and relevant design, establishing comprehensive developer workspaces and assistants, implementing quality and security gates, and setting up effective software teams. The focus should be on what genuinely generates value.”
