Track 10: Human-Centered Agentic AI
Track Chair:
1. Sidra Zafar, Kinnaird College for Women, Lahore, Pakistan
Track Co-Chair:
1. Aasia Khanum, Forman Christian College University, Lahore, Pakistan
2. Zeeshan Zia, Chief Scientist, Retrocausal, Inc. Seattle, USA
The track aims to target research on the next generation of Autonomous Agents that are able to plan, call out tools, retain memory, and interact with the world with minimal human interaction. We particularly welcome papers on pragmatic deployments in real-world domains, multi-agent coordination, adversarial security threats specific to agentic pipelines, safety and alignment mechanisms, calibrated human trust and explainability, and ethics and governance of delegating consequential decisions to AI systems.
Topics of interest include, but are not limited to:
- Agentic Systems for Real-World Domain Applications
- Multi-Agent Orchestration and Communication Architectures
- Dynamic and Adaptive Multi-Agent Systems
- Prompt Injection, Agent Hijacking and Tool Poisoning.
- Agentic Pipelines Adversarial Detection and Defense
- Safety Constraints, Guardrails and Human Oversight:
- Accountability, Regulation and Policy for Agentic AI
- Localization and Resource-Constrained Deployment of Agentic AI in Global South.
- Trust, Explainability and Human Agent Interaction
- Delegation Boundaries and Human-Agent Collaboration Models
- Benchmarking, Evaluation and Robustness of Agentic systems.
- Novel Agentic Architectures, Engineering Approaches and Design Patterns