Artificial intelligence-enabled management platforms and infrastructure are beginning to enter enterprise networks. I say “starting” because despite a lot of AI-wash marketing efforts over the past few years, a lot of the stuff that’s been described as “AI-driven” or “AI-powered” hasn’t really materialized. It’s not that these systems don’t do what marketers say they do, it’s that they don’t do what they imply.
Even if some tools do actually use AI in meaningful ways and produce results that are significantly different than what might have been produced without it, it doesn’t feel qualitatively different from what came before. They may be better, for example by significantly reducing the number of false positives in alert traffic, but they are not different.
Now that’s starting to change. Artificial intelligence tools that feel different, changing the way network administrators use their tools, are starting to enter the market. A good example is the introduction of virtual assistants that can have meaningful conversations about what is happening in the network and can (if allowed) take actions that change the functionality of the network. This shift from just another tool to a colleague of sorts will make it clear to networking teams that real change is happening in a way that spec sheets and better UI can’t. They will make it obvious that IT is entering new territory.
It’s never too early. The networking field’s population is starting to feel a little thinner and a little older, as long-time engineers and administrators retire or move to other types of jobs and aren’t replaced by hordes of eager young newcomers. Networking has never been the sexiest job, and much of the excitement in enterprise IT hasn’t been focused on networking for years. People entering technology fields are more likely to be drawn to fields like robotics, metaverse programming, data science, and (yes) artificial intelligence.
The demographics of network staff are such that any medium to large network will inevitably use AI tools in the short term as they will be perceived as easier to obtain and use than new hires. Over the next seven or eight years, any size network will The web will do the same as AI will be increasingly integrated into the platforms themselves.
Like many other forms of automation, the dynamics of AI-infused network organizations will center on four modes of interaction: offloading, retraining, deskilling, and displacement.
AI offloading means putting AI tools under the command of trained and experienced network professionals to help them do their jobs. The idea is to make networking professionals more effective by letting them offload tasks that are repetitive, complex, time-sensitive, or require a high level of concentration but not creativity. This should free up these scarce and valuable resources to do other higher-level work while giving minimal supervisory attention to what the AI is doing. (Human attention is the most valuable resource in any IT shop.) The networking team won’t shrink, and its service portfolio can even grow without the team having to grow to do so.
Reskilling allows network employees to be trained to move into other parts of IT or to do a completely different type of job. It also includes the idea of using AI to help train new network employees to proficiency. The network team may shrink or see more turnover, but its ability to get the job done will not diminish.
Deskilling is a different outcome that we saw in tool and die work after World War II. (For more information, check out David Noble’s Productivity.) New tools are introduced to allow less skilled workers to do the work of more skilled workers, without any intention or permission for them to become more skilled as they work. Entire areas of expertise will be moved to silicon and removed from the job requirements of most positions. This transfer of skills to software or firmware makes it easier for businesses to find the right networking personnel because the requirements are lower.
Replacement is the end of the deskilling spiral, and AI tools in the hands of IT generalists simply replace network experts. This could be something management is doing to cyber teams preemptively in an attempt to rid themselves of the burden and cost of staffing, or it could be something cyber teams are doing to themselves, using AI to make the organization no longer a soft landing in the ability to hire and retain. Staff are skilled enough to do the job.
Network engineers, administrators, and IT leadership already need to think and plan why and when to adopt AI tools, how to use them to best effect, and how to reinvent enterprise networks.