AI-enabled management platforms and infrastructure are beginning to make their way into enterprise networks. I say “beginning” because despite lots of AI-washing marketing efforts over the last few years, a lot of what has been characterized as “AI-driven” or “powered by AI” hasn’t really materialized. It’s not that these systems don’t do what the marketers say, so much as they don’t do it in the way they imply.
Even some tools that do truly employ AI in meaningful ways, and with visibly different results than are possible without it, don’t feel qualitatively different from what has come before. They may be better, for example by dramatically reducing the number of false positives in alert traffic, but not different.
That’s now starting to change. AI tools that feel different to work with, that change the way network admins work with their tools, are beginning to enter the market. A solid example is the introduction of virtual assistants that can have meaningful conversations about what is happening in the network, and that can (if allowed) take actions that alter network function. This shift from just-another-tool to sort-of-co-worker will make it clear to network teams that a real change is taking place in a way spec sheets and better UIs can’t. They will make it viscerally apparent that IT is entering new territory.
Not a moment too soon. The population in network-land is starting to feel a bit sparse and a bit old, as long-time engineers and admins retire or move on to other kinds of work and are not replaced by hordes of eager young newcomers. Networking has never been the sexiest of jobs, and most of the excitement in enterprise IT has not centered on networking for many years now. Folks entering tech are far more likely to be pulled to fields like robotics, metaverse programming, data science, and (yes) AI.
The demographics of network staffing being what they are, it is inevitable that any mid- to large-size network will wind up using AI-powered tools in the near term, since they will be seen to be more easily acquired and utilized than new staff. Every network of any size will do so in the next seven or eight years, as AI will increasingly be baked into the platforms themselves.
The dynamics of AI-infusing a network organization will, as with many other forms of automation, center on four modes of interaction: offloading, reskilling, deskilling, and displacing.
AI offloading means putting AI tools at the command of trained and experienced networking professionals to help them do their work. The idea is to make network pros more effective by allowing them to offload tasks that are repetitive, complex, time sensitive, or require extremely high levels of focused attention, but that are not creative. This is supposed to free these scarce and precious resources to do other, higher-level work instead, while paying minimal and supervisory attention to what the AI is doing. (Human attention is the most precious resource in any IT shop.) The network team doesn’t shrink, and its portfolio of services can even grow without the team also having to grow to make that possible.
Reskilling allows network staff to be trained to move into other parts of IT or into entirely different kinds of jobs. It also encompasses the idea of using AI to help train new network staff up to proficiency. The network team might shrink or see more turnover, but its ability to get the work done does not diminish.
Deskilling is a different kind of result, one we saw at work in tool and die work in the wake of World War II. (Check out David Noble’s Forces of Production for the details.) New tools are brought in to allow less-skilled staff to accomplish the work of more-skilled staff, and without any intention or allowance for them becoming more skilled as they work. Whole areas of expertise would shift into silicon and fall off the job requirements for most positions. This shift of skills into the software or firmware makes it easier for the enterprise to find suitable network staffers because requirements are lower.
Displacing is the endpoint of the deskilling spiral, with AI tools in the hands of IT generalists simply replacing network specialists. This might be something done to the network team preemptively, by management seeking to rid itself of the burden and cost of staffing, or it might be something done by the network team to itself, using AI to enable a soft landing for an organization no longer able to hire and retain staff skilled enough to do the work.
Network engineers and admins and IT leadership already need to be thinking about and planning around why and when to adopt AI tools, how to use them to best effect, and how to reshape enterprise networking in their wake.
Copyright © 2023 IDG Communications, Inc.