Aug 2026· IEEE Transactions on Pattern Analysis and Machine Intelligence· Vol PP, pp. 1-20· 9 citations
Medicine
TL;DR
This survey bridges the existing gap by presenting a comprehensive blueprint for scientific agents' design and introduces a unified taxonomy based on capability envelope and capability maturity, characterizing both the scope of scientific workflow coverage and the reliability of agent behavior under realistic research conditions.
Abstract
The advancement of LLM-based agents is redefining AI for Science (AI4S) by enabling autonomous scientific research. Prominent LLMs exhibited expertise across multiple domains, catalysing constructions of domain-specialised scientific agents. Nevertheless, the profound epistemic and methodological gaps between AI and the natural sciences still impede the systematic design, training, and validation of these agents. This survey bridges the existing gap by presenting a comprehensive blueprint for scientific agents' design. It first clarifies the concept of scientific agents and distinguishes them from general-purpose agents in terms of their goal orientation, workflow embedding, and scientific commitments. It then introduces a unified taxonomy based on capability envelope and capability maturity, characterizing both the scope of scientific workflow coverage and the reliability of agent behavior under realistic research conditions. Building on this taxonomy, the survey further connects scientific agent design with the research life cycle by reviewing construction strategies, capability enhancement methods, evaluation paradigms, and future challenges. This unified perspective aims to provide practical guidance for designing domain-specific scientific agents and to promote the convergence of AI research and natural scientific discovery. To support long-term progress, we curate a live repository (AWESOME_SCIENTIFIC_AGENT) that continuously aggregates emerging methods, benchmarks, and best practices.
Overall, it is found that the use of coding agents in scientific computing holds great promise for accelerating scientific research and increasing the reliability of critical systems, but that outstanding concerns remain.
Jeremiah H. Li, Alex Rubinsteyn, Sergey Feldman et al.· bioRxiv· 0 citations
This work argues that studying AI Scientists as human-agent systems (HAS) is both underexplored and undervalued, and calls for new research that adopts the HAS lens to develop mathematical frameworks for understanding and fostering human-AI synergy in scientific discovery.
P. Emami, Sameera Horawalavithana, T. Nguyễn et al.· 0 citations
A unified, taxonomy-driven, and deployment-oriented survey of agentic AI systems, synthesizing recent advances through a modular reference architecture and a four-dimensional taxonomy that characterizes agents along the axes of autonomy, tool use, collaboration, and safety–governance is presented.
Sparsh Bajoria, Shreyanshu Ranjan, Adhitya M et al.· Cognitive Computation· 0 citations
GPT-5.5-like systems and their place represent in progressing agentic artificial intelligence (AI) in library environments. This study aims to investigate how autonomous, purpose-directed AI agents can disrupt fundamental library operations, including information access, user services, knowledge organization and research support, while meeting shifting user expectations in digital knowledge ecosystems.
This research is analytical-conceptual, based on an extensive review of recent literature regarding agentic AI and smart library systems. It reviews various theoretical models and practical applications and extracts important directions, technological features and areas of function that are important in context of conventional libraries that will be enabled for AI in future.
The results suggest that advanced models like GPT-5.5can run away with agentic AI systems. It provides substantial improvements in automation, personalization and proactive assistance in library services. Systems have the ability to perform complex operations independently, including semantic search, metadata generation, user interaction and guiding the user in the research process. The issues surrounding data privacy, good governance and the transparency of such systems remain major roadblocks.
This paper has added to existing conversation about AI and libraries by adopting an agentic lens with an emphasis on capabilities and autonomy. It provides advice for researchers and practitioners aiming to conceive future-ready, smart library systems.
This comprehensive review paper examines the ontological shift in Open Source Intelligence (OSINT). For decades, the field was strictly defined by the extraction of actionable intelligence from publicly accessible data, operating on a linear, highly manual pipeline that relied entirely on human cognitive processing. The advent of Agentic Artificial Intelligence (AI) fundamentally displaces this foundational paradigm. By transitioning the discipline from reactive data aggregation to proactive, autonomous execution, agentic architectures redefine the intelligence lifecycle, shifting the cognitive burden of data processing, correlation, and initial synthesis from human operators to autonomous algorithmic systems. To clarify the academic boundaries and methodological intent of this document, it must be explicitly stated that this is a comprehensive review paper rather than a presentation of singular novel empirical research. The objective of this review is to systematically aggregate, synthesize, and critically evaluate the theoretical, mathematical, and architectural state-of-the-art across the rapidly expanding domain of Agentic AI in cybersecurity and OSINT. By analyzing recent academic frameworks published across leading repositories—such as the Institute of Electrical and Electronics Engineers (IEEE), the Association for Computing Machinery (ACM), Springer, and Elsevier—this paper provides a definitive structural analysis of the current landscape. This review specifically addresses the mathematical foundations of autonomous reasoning through Bayesian probabilistic updating, constructs structural models of multi-agent collaborative topologies, evaluates the operational efficacy of specialized cybersecurity frameworks, and thoroughly examines the profound socio-technical friction generated by this paradigm shift, specifically focusing on the barrier of algorithmic trust, legal accountability deficits, and the emerging defensive imperative to track Non-Human Identities (NHIs).
Arunanshu Chatterjee, A. Roy· Interdisciplinary Journal of...· 0 citations
Experimental results indicate that coordinated autonomous agents significantly reduce research time, improve workflow consistency, enhance knowledge discovery, and increase scientific productivity compared with conventional AI-based research assistants.
Anatoly Kitov, M. Kartsev· International Journal of Eme...· 0 citations