Comment: Engineers should be collaborators and not competitors with AI

Today, there is an expectation that engineers should understand how to use advanced tools and systems that leverage artificial intelligence (AI) to accelerate engineering tasks through automation. However, 71 per cent of modern engineers are concerned that AI will replace them.

While their concerns are valid, the future of engineering is not one of competition (humans vs machines) but one in which AI and engineers are co-collaborators. In fact, by becoming proficient with AI, modern engineers will significantly enhance their capabilities while accelerating the software development lifecycle (SDLC).

Ways AI and engineers can collaborate

AI applications, like Microsoft Copilot, empower engineering teams to automate time-consuming tasks like writing code.  For experienced engineers, Copilot can be a powerful tool to lay the groundwork for their engineering tasks.  Once code is suggested, the engineer just needs to make simple edits to ensure the code meets the expected standard.  This way of working will allow engineers to redirect the time and effort spent on monotonous work to other business objectives to generate more value.

MORE FROM ARTIFICIAL INTELLIGENCE

One of the more promising advancements in AI is the concept of Agents. According to Forrester, AI Agents, or TuringBots, can help software developers and entire development teams plan, design, build, test and deploy application code. Like a personal assistant, Agents will collaborate with humans at various stages of the SDLC, whether writing and deploying or testing and spinning up new environments, empowering engineers to truncate traditional timelines, achieving much more in considerably less time.   These Agents will empower less experienced engineers to be more productive than those not using them. 

Training engineering teams to embrace AI

Not only do enterprises need to equip their engineers with the right tools but also the right training to attain this future where engineers and AI are co-collaborators. It’s critical that teams evaluate how they are working before investing in these tools because adopting AI with an immature SDLC will result in limited benefits.  Therefore, companies must focus on change management at the task and tool levels, among individuals and teams for proper adoption and change readiness.

From a holistic perspective, enterprises can begin by establishing security guidelines and rules of engagement with AI. These standards will empower teams to experiment while minimising risk. Companies should also implement quick-win use cases wherever possible, e.g., code generation, task automation and issue analysis. Likewise, businesses can create working groups where team leaders encourage individuals to share what is and isn’t working with AI.

Companies should note that legacy teams may require additional care. Even tools like Microsoft Copilot still require change management activity. There should be an effort to provide legacy engineering staff with ample reskilling opportunities. Reward mechanisms will also incentivise the utilisation of new AI tools. The realignment and restructuring of teams may be inevitable, and companies must plan accordingly.

The necessity of prompt library platforms and embeddings

Companies can invest in a prompt library platform to enable engineers to unlock AI’s massive potential. Consider a scenario where an engineer writes useful ChatGPT or OpenAI prompts but can't share them with their team members or the larger organisation. Because the engineer can’t share their work, the value of AI remains siloed with that one individual. Alternatively, a prompt library platform will permit engineering teams to share and reuse prompts within the company, which is particularly advantageous for larger enterprises with multiple lines of business.

Prompt library platforms can also promote greater collaboration between engineers and AI by allowing engineers to create and deploy embeddings across the business. Embeddings are important because they help AI models gain contextual understanding. When an API lacks embeddings, the AI model will generate generic code that doesn’t integrate well with existing architecture. Best-in-class prompt library platforms with embeddings enable AI models to generate more suitable code, empowering engineers to eliminate tedious work.

Additional considerations for fostering collaboration

The fear of AI is real, but with the right training and support, engineers can expertly collaborate with this technology to achieve exponentially greater outcomes. Of course, there will be growing pains and resistance to change, whether conscious or unconscious.

To that end, businesses must strive to remain transparent and explain the reasoning behind AI implementations. Cross-functional collaboration between data scientists and various IT professionals will also be key to building an AI-focused enterprise. Moreover, leaders can drive AI adoption by planning for the jobs of the future rather than setting up to eliminate current ones.

Adam Auerbach, VP, DevTestSecOps Practice at EPAM Systems, Inc.