There is growing realisation among IT leaders that artificial intelligence (AI) will become a powerful tool to help them tackle the challenges of driving digitisation at pace while handling a shrinking talent pool.
Recent research has shown that AI has the potential to plug the skills gap in software development. Providers of robotic process automation (RPA) tools also see a huge opportunity in the use of AI-based code generation for speeding up RPA.
The use of AI-based software development is something many IT leaders feel is likely to become commonplace to address the growing IT skills shortage, with more talent needed to support digital business initiatives.
A 2022 Nash Squared digital leadership report, based on a global survey of 1,785 IT leaders found that the war for talent, and keeping pay demands reasonable, emerges as a big challenge. The survey reported that 70% of digital leaders state that a skills shortage prevents them from keeping up with the pace of change.
Nash Squared also found that 62% think that organisations will never have enough technology staff and a similar number (60%) feel that the rising cost of living has made salary demands unsustainable.
According to the Nash Squared survey, 74% of digital leaders feel that their government’s policies are completely ineffective at tackling the skills shortage. But many feel that AI offers an opportunity to plug the gap in the tech talent market. Nash Squared reported that, on average, digital leaders plan to use digital labour to automate around 1 in 6 (16%) of their workforce over the next five years.
The viability of AI software developers
In August, researchers published a paper, The communicative agents for software development, which demonstrated the viability of using AI to automate programming.
The researchers said their experimental results demonstrated the efficiency and cost-effectiveness of the automated software development process, which was driven by ChatDev, a large language model-based tool for creating customised software using natural language.
“By employing multiple agents with different roles, we have proposed a new paradigm in generating software systems, alleviating code vulnerabilities, and identifying and resolving potential bugs,” they stated in the paper.
Although the study may potentially help junior programmers or engineers in the real world, the researchers found that it is challenging for the system to generate perfect source code for high-level or large-scale software requirements. They said that this difficulty arises from the AI agents’ limited ability to autonomously determine specific implementation details, often resulting in multiple rounds of lengthy discussions.
Additionally, large-scale software development proves challenging for both AI-base code reviewers and testers, as it becomes difficult to identify defects or vulnerabilities within the given time constraints.
AI in low code
Gartner’s latest Magic Quadrant report on RPA found that almost every RPA provider has either already experimented, or plans to experiment, with including GPT models in their products or enabling APIs to known GPT providers such as OpenAI.
The report’s authors found that RPA companies are also offering direct capabilities, such as automatic email generation, prompts within workflow design, more accurate intelligent document processing model training, process-mining analytics and AI-driven development of RPA scripts. Gartner believes that many RPA providers will launch incremental capabilities in the next one or two years, making AI-generated automation an important capability.
Discussing the skills crisis and how IT leaders need more and more software developer time to build software that supports new digital business initiatives, Jesper Hyrm, international vice president, GitHub, said: “We can’t keep bootstrapping developers and expecting them to work harder in the pursuit of innovation. Something has to give if developers are expected to keep up with both spiking demand for their services and the skyrocketing pace of technological change.”
Hyrm believes AI-based software development tools will play a pivotal role in easing the developer shortage and allowing enterprises to innovate at speed. “Generative AI signals a sea change in software development that is turbocharging developer productivity and firing creativity among developers,” he said.
According to Hyrm, the use of AI tools that eliminate the drudgery of repetitive tasks allows developers to stay “in the flow” longer – freeing them to focus on more impactful work rather than having to constantly switch environments and spend time researching how to write the code.
“Developers can focus all their creativity on the big picture: building the innovation of tomorrow and accelerating human progress. Reducing boilerplate and manual tasks and making complex work easier allows developers to become the best versions of themselves, producing their best work,” he said.
While the experiments with AI-based software development have highlighted limitations of the technology, there is little doubt that it will mature to a level where IT leaders will call on AI-based developers to extend their internal teams.