There were system administrators. They conducted backups, designed VPNs and created disaster recovery plans (DRPs). Managers didn’t understand them and thought they were barely working. For a while, everything was calm and good.
Then came DevOps. The development immediately sped up, pipelines started working better, products had fewer bugs and the developers were finally happy. For a while, everything was good again.
Then the containers arrived. Everyone rushed into Kubernetes as if the train had already left the station. They practiced Kubernetes for the sake of Kubernetes, for the buzzword, and the chance to show they were at the cutting edge. For a while, everything was good again.
And then artificial intelligence (AI) came — a revolution, no less. Hysteria ensued in the IT sector about how nobody would be needed in the foreseeable future as everything would be done by AI. But before we start panicking, let’s figure out what AI does well, where it falls short in DevOps, and whether DevOps professionals should fear it and start re-skilling today.
Where AI Excels and Where It Stumbles
AI is a superb helper in automating repetitive tasks and an excellent assistant in development. It checks syntax, makes code beautiful, inserts functions, writes code comments, finds bugs you might have missed, succeeds in testing and generating tests and performs well in log analysis. It also helps predict system behavior, loads, and issues and works well in detecting threats and responding to events.
However, AI lacks a deep understanding of context, critical analysis, creativity and motivation — crucial aspects of DevOps, which is fundamentally about communications. AI also struggles with building complex processes and interactions between diverse and intricate parts of a system.
Where DevOps outshines AI
DevOps is a practice that promotes collaboration between software developers and IT professionals while automating software delivery and infrastructure changes. Its goal is to create a culture and environment where development, testing, and releasing software can happen rapidly, frequently and reliably. AI is not yet the magic wand that can solve human-centric tasks.
DevOps extends beyond pipelines, servers, Kubernetes and Docker, emphasizing teamwork and strong relations within a team. With DevOps, no separate teams are sticking to their specific rules; instead, there is a unified delivery team where everyone is responsible for issues and eager to solve them. DevOps also involves inter-departmental communication, knowledge transfer, and interaction during deployments.
What About the Technical Side?
Naturally, AI can assist you in writing scripts in Bash, Python or PowerShell — ask ChatGPT and it will be provided along with comments. However, thinking through a complete DRP with testing and backup verification is beyond AI capabilities. How do you explain to it that the tapes we use for backups need to be stored in a bank and the data recovery process should be tested quarterly, potentially failing due to a miswritten phone number?
Or for instance, take containers and Kubernetes. Precise configurations of your deployments require a lot of work — PSP, PodDisruptionBudget, good limits and requests, solid probes (set up so they don’t crash the cluster), security configuration and proper monitoring. Relying on AI for these tasks is impractical — you need a DevOps engineer competent to implement them.
AI vs. DevOps: Who is the Winner?
How often have you written a script and felt too lazy to add comments or write a README? How many times have you manually searched for a bug in the code and fixed it (line by line, through long nights)? How frequently have you had to wake up at night due to a production notification? This is where AI comes in, automating those and similar tasks.
There will be no battle for survival in the future. The strength of AI is unleashed when combined with a human-centric approach. That means, DevOps will plan, execute, control, and improve, while AI will help handle tasks faster and with greater accuracy.
Good Specialists Learn Continuously
DevOps must improve their knowledge, hone skills and learn to handle conflict situations — not just because of perceived AI threats, but to become and remain in-demand specialists. If you are flexible, easily adapt to changes and face challenges head-on without lamenting over the past, no AI can replace you. AI will assist you, asking, “How’s it going today? What can we improve and accomplish today?”
