The Robots Are Coming (But Not Yet), or, Don't Make Smart People Do Dumb Things
summary
Argues that AIOps today detects and alerts but rarely remediates autonomously, using ML to correlate Prometheus, Kubernetes and OpenStack logs and predict failures after a six to twelve week training on an environment, not a plug‑and‑play tool. Cites intrusion detection models that correlate traffic patterns and can throttle or cut offending sources in real time, warns about need for vast historical data, distributed learning (projects like Flower), and human approval gates because true automated remediation isn’t reliable yet.
AI-written summary · about
chapters (22)
- 0:00Intro
- 0:30Who am I
- 0:59Todays topic
- 1:36Current scenario
- 2:16Challenges
- 2:56Keeping staff available
- 4:00The dream vision
- 5:09How is the decision made
- 5:30How do we teach machines to do that
- 6:13AI vs ML
- 7:42What is AI
- 8:42What is Machine Learning
- 9:25Compounding Information
- 10:34Machine Learning
- 12:24The Real World
- 14:09Use Cases
- 19:35Limitations
- 21:32The Problem
- 22:38Biases
- 24:53Blue Sky Future
- 25:10Intelligent Automation
- 27:08Final Thoughts
related talks
34mWhat If the Promise of AIOps Was True?
20mAutomated Incident Management
05mAutomating Cloud-Native Infra with Intelligent Tools
33mHuman Factors in the Age of AI Ops: Re-Engineering Trust between Humans and...
09mAI/ML Won't Save you
29mAutomation first, People centric
24mAIOps, are we there yet?
27mAI-Powered Reliability: Stop Fighting Fires, Prevent Them