- 12 formas de acelerar la transformación digital
- EU Commission to Invest €1.3bn in Cybersecurity and AI
- The 160+ best Amazon Spring Sale tech deals live
- Federal Desktop Core Configuration (FDCC/USGCB) Compliance
- Ana Peñuela (KPMG): “Buscamos convertir al CIO en el 'partner' estratégico de los negocios para impulsar la digitalización”
Network Engineers: I Don't Want to Be an AI Babysitter

I get more excited every day as I learn something new. However, I also have my fair share of concerns about the future—specifically on the topic of AI and how it will impact the role of network engineers. Okay… I probably have more than my fair share of concerns. (That won’t come as a surprise if you’ve been following the last few years of my journey, exploring the “AI FUTURE!!!”)
First off, I want to be very clear. I’m excited about the future of network engineering, network automation, and my place in this wonderful world and community. In fact, my recent blog, Navigating the AI Era as a CCIE, discusses how awesome it is to be a CCIE right now.
I generally focus on where I see the positive possibilities. How AI can make our lives and work as network engineers better.
But today, I want to talk about something that worries me: how the AI future is being discussed and described. My hope is that by discussing it, we can avoid the worst possible dystopian vision of that future. While I like reading books or watching movies about these dystopian futures (a guilty pleasure of mine), I don’t want to live in one of those worlds. I’m also hoping that you, my community, can help me understand whether my concern about the future of AI is overblown. So, let’s dive in, shall we?
I don’t want to be an AI babysitter…
There is a phrase that has been showing up in presentations, blogs, articles, videos, press releases, government documentation, and just about everywhere else discussing how AI will impact the future of work. The phrase refers to an approach called “human-in-the-loop.”
So, what is “human-in-the-loop?”
I just did a Google search for “‘human in the loop’ ai cisco” and Gemini was helpful in giving me this summary:
Cisco emphasizes “human-in-the-loop” AI, meaning integrating human oversight and feedback into AI systems to ensure accountability, ethical considerations, and reliable decision-making, especially in areas like security and data analysis.
That doesn’t sound bad, right? Here’s another snippet from a paper I recently read on AI and the future of job roles:
The extent to which it [Gen AI] can replace humans in the workplace will depend on the necessity for human oversight of machine-performed tasks.
No doubt you’ve seen or heard similar descriptions of what it will take to “safely” integrate AI into day-to-day tasks. Here’s my understanding of why human-in-the-loop comes up over and over in discussions.
It comes down to a few points:
- Using AI offers a “value” businesses can NOT ignore. What that value is can vary, but it generally comes down to speed: AI is simply faster than humans.
- AI isn’t always right. And AI can’t be held accountable for mistakes.
- By having a human sign off on the AI work, mistakes will be caught. And if they aren’t, there is someone to be held accountable.
I’m NOT saying that the above points are factually valid. In fact, each of those statements on their own deserves a lot of deep consideration and discussion. But for the sake of this blog post, let’s take them as they sit to further explore my concerns about a future where Hank is a “human in the loop” for AI systems.
Here’s the problem with “human-in-the-loop”
I like being a network engineer. I like creating network designs to meet business demands. I enjoy creating configurations and engineering robust routing protocols. I find the process of troubleshooting a network issue rewarding.
I’ve spent years of my life learning the skills it takes to DO network engineering. And I still have many years ahead of me as a network engineer. I also have a lot to offer the companies, networks, and team members I will work with in the future.
Every description I’ve read or heard about “human in the loop” places the human near or at the end of “the loop.” An AI tool is posed a problem, question, or set of data to work on. Then, AI generates its solution, which is then sent to a human to review, accept, reject, or make changes.
When I think about this concept, I can’t help but conjure up a picture of row after row of humans spending their days listening for the “ding” of a new proposed AI work item, waiting for the human to do their thing so the AI can continue on its “loop,” completing the work. That just doesn’t sound like the future network engineer I want to be.
Which will come first: AI or experience?
There is something else I wonder about in this “human in the loop” vision of the future. A human network engineer’s ability to identify a mistake made by AI relies on whether that network engineer has made that same mistake in the past. Or, at the very least, they need enough network engineering experience to notice when something is wrong.
As of now, we have experienced network engineers who can “oversee” AI agents and identify potential issues. Heck, that’s half of what senior network engineers and CCIEs do anyway: support the up-and-coming network engineers on our team by reviewing their work and helping them learn from their mistakes.
But how will future up-and-coming network engineers gain the experience of being a network engineer if they are simply a cog in “the loop?”
And yes, I am fully aware that this is an extreme example and not what people mean when they say “human in the loop” or “human oversight.” Regardless, it is critical that we consider this type of extreme outcome now, when the future of network engineering is being written. Because I absolutely think there is a way this narrative can be turned around—a future vision where network engineers continue to be network engineers more than in name only.
Let’s turn it around: “AI-in-the-loop”
I propose that we invert the loop. Make no mistake—artificial intelligence absolutely offers value to network engineers doing network engineering jobs day in and day out. In fact, I use it myself. But I use AI as a resource—like any other—at my disposal.
Suppose I’m called in to troubleshoot an intermittent routing problem at our Internet edge. Using my well-worn network troubleshooting skills, I gather details about the issue, perform different tests, and try to replicate it. I inspect operational output from the routers and look at our network management systems. Maybe I ask around, “What changed?”
And if everyone tells me, “Nothing. Nothing changed.” I then ask, “Well, what changed before nothing changed?”
As I do all of this, I leverage many tools and resources. I’ll consult our internal documentation about the network. I’ll review the recent change requests. I might head over to Cisco.com and search for error messages or scenarios. (Well… no, I’ll probably go to my favorite search engine and search for error messages and scenarios. 🙂 )
It is here, during this part of my work, where I’ll bring AI into “the loop.” Not only is AI fast, but it has been trained on and has instant access to all sorts of useful data that is relevant to my work.
AI-in-the-loop: A tool for network engineers
I may be struggling to remember the exact show command to display all the details about the BGP prefixes learned by my router. Or I may want to set up a filtered packet capture and am looking for an example configuration. Or I’m reviewing hundreds of lines of debug messages and could use help in quickly finding the anomalies. These are examples where AI can make ME a better, more efficient network engineer.
You see, I am a network engineer. I’m a pretty decent network engineer. I’ve typed millions of CLI commands with my fingers, seen countless pings drop, configured routing protocols, access control lists, VPNs, policy maps, EtherChannels, and so on and so on. But I’m still just a human, not a computer. I may not have instant access to everything buried in my brain, but I know when the answer is in there. I know that if I see the correct answer (or something close), I can recognize it and get to the solution. It’s the same reason an experienced network engineer can solve a complex problem with one web search and a glance at a forum post or Cisco command reference.
We should stay in the driver’s seat. We should stay in control of the networks and the network engineering. We should embrace the capabilities of AI to improve our network engineering work. AI shouldn’t be using us to improve its network engineering work—we should be using AI as a resource to become more effective network engineers—now and into the future.
Really Hank… is that all AI should be?
So, you might be thinking:
Oh, Hank, you nice old boomer network engineer. Get with the times… AI offers us way more than just a next-generation search engine!
Yes, it absolutely does—and I’m excited about a lot of the enhancements to the systems and software we use every day. Not to mention the completely new systems and software that are enabled by AI. Just looking at Cisco’s announcements in the AI space this past year excited me about its potential for network engineers.
Just imagine what we’ll be able to do in the future. Since the first network engineer started capturing log data, we’ve recognized that it’s nearly impossible for a human engineer to make sense of the flood of information in any timely fashion. Think of all the outages that could have been prevented if we were able to find the small and early hints buried in counters, NetFlow data, and log details. As for security… wow. There is so much potential in the security space to identify and respond faster.
Embedding AI capabilities into networking products will give us a massive boost as network engineers. But this also isn’t anything all that new. For many years now, machine learning capabilities have been added and iterated on to enhance the network assurance solutions for the campus, WAN, and data center. They are getting a new boost from the GenAI hype and buzz right now, but most of them aren’t GenAI.
Something is coming to the network engineers’ world that relates to GenAI that has me very, very excited. Natural Language Interface, or NLI, will soon join the much loved and lauded Command Line Interface (CLI) and the slightly-bummed-it-isn’t-the-new-kid-on-the-block-anymore Application Programming Interface (API) as methods network engineers interact with the devices and systems we manage. And that will be awesome. Truly, a game changer.
Yes, part of becoming a network engineer is learning all the specific commands required to make the network work. When network engineers gather together and share war stories, someone will always complain (lovingly) about how it makes no sense that it is “ip ospf authentication-key” but “ip authentication mode eigrp,” and why can’t they just be the same?! And we’ll laugh and laugh and laugh.
But let’s be honest. It isn’t memorizing specific command line syntax that makes us network engineers. It is knowing how, why, and when we need to configure authentication for our routing protocol that is important. Won’t we be so much happier when we can simply tell our router:
“Enable authentication for EIGRP and OSPF on all interfaces. EIGRP should use md5 with key-chain 5, and OSPF needs to use plaintext because of the legacy device we are connected to.”
Sure, some network engineers will grumble and say things like “back in my day.” But I know I’ll be happier for it all.
So what now?
So what now, you ask? Well, I want to hear what you all think. Don’t be shy. If you think I’m overreacting, please tell me. If you share my concerns, let me know I’m not alone. What excites you about the future of network engineering with an AI assistant in your pocket? Are there some tasks you can’t wait for AI to take over for you? Leave a comment below to let me know your thoughts!
In the meantime, here are some suggestions for excellent places to learn more about AI and start building skills. Because there is one thing I’m absolutely sure of… AI is coming, and we gotta be ready for it.
- Spend about 45 minutes Understanding AI and LLMs as a Network Engineer with this great tutorial by Kareem Iskander.
- Invest more time in this excellent Network Academy course, Introduction to Modern AI, with my new favorite instructor, Eddy Shyu. (Don’t let the fact that it’s on Network Academy scare you away. It is fantastic for anyone looking to get a solid foundation in AI.)
- Dive in deep and “Rev Up” your recertification journey (34 Continuing Education credits!) with AI Solutions on Cisco Infrastructure Essentials. Free in Cisco U. until April 26, 2025, and with content and videos from 5xCCIE (and my hero) Ahmed Moftah.
Sign up for Cisco U. | Join the Cisco Learning Network.
Follow Cisco Learning & Certifications
X | Threads | Facebook | LinkedIn | Instagram | YouTube
Use #CiscoU and #CiscoCert to join the conversation.
Share: