We Can’t Trust Networks to Fix Themselves

AI is changing the way we maintain and troubleshoot networks, offering suggested fixes and knowledge about issues. But we aren’t ready to trust it in fully autonomous mode yet. In this episode of the Tech Field Day Podcast, Dan Ryan, Chris Reed, and Alan Wang join Tom Hollingsworth on-site at Mobility Field Day to discuss why fully autonomous networking still feels out of reach despite rapid advances in AI. The group explores the tension between deterministic networking protocols like OSPF and BGP and the non-deterministic nature of LLMs, noting that while AI excels at parsing data and suggesting fixes, it often lacks the business context needed to distinguish real problems from expected anomalies. They also raise concerns about a growing skills gap, arguing that overreliance on AI for Tier 1 and Tier 2 troubleshooting could prevent junior engineers from developing the deep expertise needed to validate AI-generated recommendations. Ultimately, the delegates agree that networking still requires a human-in-the-loop approach, with AI best suited for guided troubleshooting, scripting, and low-risk operational tasks rather than fully autonomous control.