Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-0461
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-07-20T11:05:50.602273Z
status: published
lens: g_space
sector: science
headline: A survey on large language model based autonomous agents
dek: Abstract Autonomous agents have long been a research focus in academic and industry communities. Previous research often focuses on training agents with limited knowledge within isolated environments, which diverges significantly from human learning processes, and makes the agents hard to achieve human-like decisions. Recently, through the acquisition of vast amounts of Web knowledge, large language models (LLMs) have shown potential in human-level intelligence, leading to a surge in research on LLM-based autono…
gain_title: LLM-based autonomous agents leveraging vast Web knowledge show potential for human-level intelligence and enable applications across social science, natural science, and engineering.
problem_title: (none)
trace_subject: (none)
gain_reading: LLM-based autonomous agents leveraging vast Web knowledge show potential for human-level intelligence and enable applications across social science, natural science, and engineering.
gain_evidence: large language models (LLMs) have shown potential in human-level intelligence | diverse applications of LLM-based autonomous agents in social science, natural science, and engineering
problem_reading: (none)
problem_evidence: (none)
quick_read: Published March 22, 2024, this peer-reviewed survey examines the shift from agents trained with limited knowledge in isolated environments to agents built on large language models trained on vast Web knowledge. The authors propose a unified construction framework and systematically review applications and evaluation methods.

The work matters because it consolidates a fast-growing research area around LLM-based autonomy, highlighting potential for human-level intelligence while acknowledging prior limitations in achieving human-like decisions. What remains uncertain from the abstract alone is which specific applications have demonstrated robust real-world performance and what evaluation gaps persist.
limitation: 
tag: Evidence-backed gain
key_points: Paper presents comprehensive survey of LLM-based autonomous agents from holistic perspective. | Authors propose unified framework for construction of LLM-based autonomous agents encompassing much previous work. | Survey covers evaluation strategies commonly used for LLM-based autonomous agents. | Abstract notes prior agents trained with limited knowledge in isolated environments diverge from human learning.
rundown: The survey organizes literature around how LLM-based agents are built, proposing a unified framework that encompasses much of previous work on autonomous agents.

It then reviews diverse applications in social science, natural science, and engineering, and summarizes evaluation strategies, ending with challenges and future directions for the field.
sources:
- peer_reviewed | Frontiers of Computer Science | https://doi.org/10.1007/s11704-024-40231-1 | 2024-03-22
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

c696d2d4c7d9212b6c80e3666213cab08a71d192d124cf0483e17a6c6ba17f68

previous

0000000000000000000000000000000000000000000000000000000000000000