A good friend of mine is an electrical engineer for a well-known passenger train company in the United States, and we recently talked about how his job, his entire career, has taken a turn from traditional electrical engineering to computer systems integrations and building telemetry pipelines. In other words, as the industry progressed, he wasn’t replaced. Instead, his role evolved.
That probably isn’t a groundbreaking revelation considering almost every industry has been impacted by some form of digitization. Things change, and people adapt. However, what really intrigued me was the new initiative to use AI to augment the train management system.
The idea is to provide the on-board train computer as much telemetry from the train systems and environmental information like weather, track conditions, and traffic, so that it can make autonomous decisions about speed, braking, failure prediction, and so on. With sufficient information and the proper data analysis, the computer can determine what the best combination of speed and braking should be, automatically alert a remote technician, and choose the perfect amount of time to stay at a train stop to avoid stopping due to track congestion (it’s less expensive and better for the train to keep moving slowly than it is to move fast and have to stop and start again).
The premise is that the process of the human engineer manually adjusting settings and configurations based on the domain knowledge in his head and the telemetry on the screen (and through the window) is slow and error-prone. An intelligent train management system would do the same activity faster and more accurately.
I asked my friend if they were headed toward completely autonomous trains with no engineers on board. He explained that there will always be a person in the cab, and the initiative wasn’t necessarily meant to eliminate engineer positions. However, engineers’ role will change.
Instead of actively interpreting various telemetry and adjusting train settings, they would supervise the train management system and be the human fail-safe in case of emergencies. The presupposition is that as sophisticated as they’ve become, AI systems still struggle with complex, unpredictable conditions that require human intuition and deep troubleshooting experience.
If you work in networking, or really in IT operations in general, this sounds familiar, doesn’t it? Collect telemetry, analyze the data, and make informed decisions to ensure uptime, reduce MTTR, improve reliability, reduce cost, and so on.
Like modern train systems that rely on sensors, telemetry, and automation to optimize operations, network operations teams will increasingly depend on AI-driven workflows to improve performance and reduce downtime. But just as trains still require human oversight, the use of AI systems in NetOps won’t eliminate the need for skilled engineers – it will simply change their role.
I believe that’s where we’re headed in the coming years with AI agents autonomously doing many of the tasks an engineer would do. Admittedly, I’ve been in this industry long enough to have seen SDN systems come and go with the promise of self-healing networks and autonomous remediation. But I do wonder if this next iteration of incorporating machine intelligence into network operations will be a big step in that direction.
In the train industry, the AI system replaces a unique activity that the legacy engineer normally did in the cab en route. With networking, I see it a little differently. Instead of reducing the number of highly skilled network engineers or replacing one very specific type of engineer, I see AI agents first replacing a lot of the varied entry level work like gathering data, configuring interfaces, routing tickets, and so on. Like a team of level 1 NOC engineers, the AI agents would perform the daily operational duties that a senior engineer would supervise.
I expect that in time agents will perform more sophisticated tasks, so while AI will initially replace repetitive level 1 tasks, its capabilities may expand into more complex areas such as network optimization, automated troubleshooting, and even architectural decision-making. So in the same way the train engineer’s role is changing, network engineers who embrace AI tools will shift toward designing, implementing and managing these AI systems in the context of network operations.
Based on what several networking vendors are doing, as well as the experimentation I and others in the community are working on, I believe this year and in the next few years we’ll see the development of many AI agent applications and tools to perform low level networking operations work. Will that mean the reduction of level 1 networking staff? Maybe, but not immediately. However, I do believe that we will see an evolutionary shift in the role of the network engineer.
Much like how train engineers are transitioning into supervisory roles, network engineers will likely oversee AI agents that handle the more routine, operational tasks. But this shift doesn’t mean the end of human expertise. The fact is, we’ll still need to build, manage, supervise, and tweak this new team of agents in the context of networking, meaning even in a supervisory role, the human engineer needs to have a strong foundation of networking concepts and workflows.
So, will AI change the nature of network operations? Absolutely. But rather than replacing engineers, it will likely augment their capabilities, much like automation has done in countless other industries. Whether we’re talking about trains or networks, the future belongs to those who can adapt, learn new skills, and embrace the shift toward machine-assisted operations.
And while I remain skeptical of tech hype, I have to admit that this time I’m totally on-board.
Thanks,
Phil
{networkphil} 
Leave a comment