Explainable AI in 6G O-RAN: A Tutorial and Survey on Architecture, Use Cases, Challenges, and Future Research

Trust is a critical barrier to AI adoption in network management: The paper identifies that human operators' inability to understand AI/ML decision-making processes in RAN (Radio Access Network) systems limits their willingness to trust and deploy these autonomous systems, highlighting that explainability is essential for real-world implementation.

XAI methods can bridge the human-machine gap in telecommunications: Explainable AI techniques are presented as solutions to help human operators better understand how AI systems make network management decisions, reducing what the authors call "the human-to-machine barrier" in 6G network operations.

Explainability requires systematic integration into network architecture: The paper emphasizes that XAI cannot be an afterthought but must be integrated into the O-RAN (Open RAN) architecture's core design, including its RAN Intelligent Controllers, suggesting that responsible AI deployment requires architectural planning from the ground up.

Security considerations accompany explainability efforts: The survey identifies that automating XAI pipelines in network systems introduces underlying security aspects that must be addressed, indicating that transparency and security are interconnected concerns in responsible AI deployment.

Multiple stakeholders and standards are addressing this challenge: The paper reviews various projects and standards tackling XAI in networks, suggesting that responsible AI in telecommunications is becoming a collaborative, industry-wide concern rather than an isolated technical problem.