A survey of 1,129 developers and 50 project managers suggests that individual developers are significantly further ahead when it comes to embracing AI than managers of application development projects.
Conducted by BairesDev, a provider of software development services, the application developers surveyed said they are now saving on average 7.3 hours of work per week, or roughly an entire working day, with AI being used to create 25% of the code they develop.
Primary AI benefits include faster coding (65%), accelerated learning (48%), increased productivity (45%), faster prototyping (34%) and better code quality (27%), the survey finds.
Additionally, a full 88% also said they now view AI as a gateway to creating new career opportunities, with 76% believing AI makes their work more fulfilling. Key opportunities identified include automating grunt work to free up time for architecture and problem-solving (62%), specializing in AI/machine learning (45%), and prompt engineering (44%).
Most of those developers are currently using on-the-job training (66%), YouTube (58%) and paid online courses (48%) as the top ways they acquire new AI skills. Only 15% said they use formal certification programs as a way to upskill versus self-learning. However, 44% said they would like to receive AI- and machine learning-related training.
BairesDev CTO Justice Erolin said that as the pace of technological change accelerates, a hands-on, proactive mindset becomes increasingly essential. However, real progress occurs when this self-directed learning is paired with structured upskilling and reskilling programs that provide depth, validation, and long-term growth, he added.
Unfortunately, project managers are not as far along when it comes to embracing AI. The top three challenges identified are gaps in business/context knowledge (43%), shortage of AI/ML specialists (41%) and lack of internal upskilling programs (38%). Only 15% of project managers report having structured AI upskilling programs.
Project managers also noted there is a lack of clarity about the actual return on investment (ROI) clarity (38%), concerns around data privacy and security (38%) and limited resources (36%).
It’s still early days so far as adoption of AI in software engineering is concerned but it’s now clearly a matter of the degree to which it will be applied rather than if. The core issue is determining what level of faith to place in the output generated by these tools, especially as it pertains to potential vulnerabilities and the amount of technical debt that might be incurred because of how verbose the code is.
Nevertheless, it’s also apparent that AI coding is only going to continue to improve and there are plenty of use cases where the code being generated sufficiently meets the immediate need at hand. Regardless of the use case, however, that code needs to be reviewed before being included into any production environment to reduce the number of potential issues that might be encountered later on.
In the meantime, leaders of DevOps teams would be well-advised to gain some hands-on AI experience of their own to better understand what type of additional training might be needed. After all, in the age of AI it’s simply not going to be possible to lead from the rear given the rapid pace of change occurring now on an almost daily basis.
