Evaluating and regulating agentic AI: A study of benchmarks, metrics, and regulation

Оценка и регулирование агентного ИИ: исследование бенчмарков, метрик и регулирования
Athanasios V. Vasilakos, Shaina Raza, Christos Emmanouilidis, Azib Farooq, Nazmul Karim, Hasan Iqbal
2026-05-09

AI evaluationAI governanceagentic AIbenchmarks and metricsmulti-agent systems
Agentic AI represents a new generation of Artificial Intelligence (AI) systems capable of perceiving, reasoning, planning, and acting toward achieving goals with a degree of autonomy. Unlike traditional AI models that merely generate outputs, these systems maintain memory, interact with their environment, and adapt over time. However, evaluating such interactive and evolving behavior remains a significant challenge. While several recent surveys have examined agentic AI architectures, components, and applications, few have systematically reviewed their evaluation, particularly regarding performance, reliability, and governance across an evolving agentic AI ecosystem. This paper addresses that gap by reviewing recent progress in the development and assessment of agentic AI, focusing on three core dimensions: benchmarks, metrics, and governance. We analyze how current evaluation frameworks capture reasoning, planning, collaboration, and ethical alignment across single-and multi-agent systems. Ultimately, this study aims to establish a unified foundation for building trustworthy, auditable, and human-aligned AI agents. The project webpage is available at project link .
1
Agentic AI systems differ from traditional models by autonomously perceiving, reasoning, planning, acting, maintaining memory, interacting with environments, and adapting over time.
2
Evaluating agentic AI remains difficult because its behavior is interactive, evolving, and goal-directed rather than limited to static output generation.
3
Existing evaluation frameworks are examined for their ability to measure reasoning, planning, collaboration, and ethical alignment in single- and multi-agent systems.
4
The paper systematically reviews agentic AI evaluation across three dimensions: benchmarks, metrics, and governance, addressing a gap in prior surveys.
5
The study seeks to establish a unified foundation for trustworthy, auditable, and human-aligned agentic AI.
6
Оценка агентного ИИ остается сложной, поскольку его поведение является интерактивным, развивающимся и целенаправленным, а не ограничивается генерацией статических ответов.

agentic AI systems

their evaluation and governance, including benchmarks, metrics, performance, reliability, reasoning, planning, collaboration, and ethical alignment

Publication Details
Publication Date
2026-05-09
Journal
Publisher
ISSN
Cited by
7
Access Type
Author Information
Authors
Athanasios V. Vasilakos
Shaina Raza
Christos Emmanouilidis
Azib Farooq
Nazmul Karim
Hasan Iqbal
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%