Agentic AI for Equitable and Sustainable Telecom
summary
Claim: automated RL agents trained on 10 years of operator data across ~500 edge nodes can amplify historical bias, interpreting low rural usage as low demand and widening coverage gaps; adding coverage equity thresholds as hard RL constraints and running geographic fairness audits yields a reported 23% rural coverage increase while preserving urban efficiency. The talk shows using digital twins and multi‑agent coordination for energy savings but warns training and inference can consume electricity comparable to small cities so embed carbon reporting in CICD and immutable audit trails.
AI-written summary · about
chapters (12)
- 0:00Welcome and Stakes
- 0:36From 5G to 6G Autonomy
- 1:17Four Topics Overview
- 1:59Connectivity Divide Risks
- 3:01Bias at Cloud Scale
- 3:58Redlining Example and Fixes
- 5:04Sustainability Paradox
- 6:40Governance Built In
- 7:05Four Pillars Framework
- 8:20Embedding Fairness in Practice
- 9:26Governance as Innovation
- 11:13Closing Takeaways and Thanks
related talks
47mBridging the Gap To Autonomous AI-Native Telco
14mReliable Agentic AI, the Open Way
19mEthical Bias
31mFrom Single AI to Multi AI Systems
44mAI, Corporate Responsibility & Democratic Legitimacy
21mDemocratizing AI with Open Source Multi-Agent Systems
18mNext-Generation AI Agents: Orchestrating Technological Evolution
01h 01mKeynote Panel: Future of AI agents