A survey of 1,063 developers published today finds nearly all (99%) are using coding tools infused with artificial intelligence (AI) capabilities, but a third (33%) have since identified a lack of a standardized AI development processes and the lack of an ethical and trusted AI lifecycle that ensures transparency and traceability of data as the top two challenges they are now encountering.
Conducted by Morning Consult on behalf of IBM, the survey also identifies customization (32%), rate of change (31%) and infrastructure complexity (29%) as additional major challenges application developers are facing.
Maryam Ashoori, senior director of product management for watsonx.ai at IBM, said the survey makes it apparent that as application development in the age of AI continues to evolve, it’s clear that developers will need to be supported by backend software engineering teams that have AI expertise. It’s not reasonable, for example, to expect developers, on their own, to be able to determine which AI models at any given time might provide the highest level of performance at the lowest cost possible, she noted.
In addition to cost and performance concerns, organizations will also need to consider the impact AI models might have on sustainability goals as well, said Ashoori.
In fact, there may be a need for a software engineering teams to swap out one AI model for another as requirements change or advances are made using other AI models that might provide higher levels of optimization, she added.
Successful organizations are investing in the software engineering and machine learning operations (MLOps) expertise needed to support developers as they embrace AI technologies, said Ashoori. It’s simply not enough to present developers with an application programming interface (API) through which they might hopefully be able to invoke some type of AI tool, she noted. Instead, organizations should be developing templates that make it simpler for development teams to invoke AI capabilities, said Ashoori.
Not surprisingly, less than one quarter (24%) of application developers surveyed ranked themselves as “experts” in generative AI. Nevertheless, almost all (99%) are also exploring or developing AI agents. The top AI agent concern identified is trustworthiness (31%).
In general, the survey finds 41% of respondents reporting that AI tools are saving them one to two hours per day, with 22% saving three hours or more per day. Developers typically evaluate AI tools based on performance (42%), flexibility (41%), ease of use (40%), and integration (36%), the survey finds.
A total of 72% of respondents report using between five and 15 tools to create an AI enterprise application, with 135 using 15 or more tools. Only one-third of those surveyed, however, are willing to invest more than two hours in learning a new AI development tool.
While application developers are clearly spending less time coding, it’s not clear to what degree AI is actually accelerating the pace at which organizations are deploying software faster. There are multiple software engineering processes that collectively limit the rate at which most organizations can practically deploy software in production environments. The challenge and opportunity heading into 2025 will be to operationalize AI in a way that goes well beyond simply writing more code faster.
