Saved in:
Bibliographic Details
Main Authors: Tian, Qiuyu, Liu, Zequn, Xia, Yingce, Yin, Haojie, Kong, Youyong
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2606.00644
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916070537822208
author Tian, Qiuyu
Liu, Zequn
Xia, Yingce
Yin, Haojie
Kong, Youyong
author_facet Tian, Qiuyu
Liu, Zequn
Xia, Yingce
Yin, Haojie
Kong, Youyong
contents AI research often requires decisions before future evidence exists: which bottleneck to attack, which direction to pursue, or where a project should be positioned. We introduce ForeSci, a temporally controlled benchmark for evaluating whether LLM agents can make such forward-looking research judgements from historical evidence. ForeSci contains 500 tasks across four fast-moving AI domains and four decision families. Each task is paired with a cutoff-aligned offline knowledge base; post-cutoff papers are hidden during generation and used only for validation. To avoid random future-event prediction, tasks are derived from pre-cutoff taxonomy branches and evidence signals, and answer-generation backbones are selected to precede the task cutoffs. We evaluate native LLMs, Hybrid RAG, and three research-agent adaptations across four backbones. Results show that explicit evidence organization improves traceability and factual support, but gains depend strongly on the decision family. Diagnostics reveal a recurring evidence-decision decoupling: agents may cite relevant evidence while forecasting the wrong research object. ForeSci turns forward-looking AI research judgement into a controlled benchmark for evaluating research agents as decision-making systems.
format Preprint
id arxiv_https___arxiv_org_abs_2606_00644
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle ForeSci: Evaluating LLM Agents for Forward-Looking AI Research Judgment
Tian, Qiuyu
Liu, Zequn
Xia, Yingce
Yin, Haojie
Kong, Youyong
Artificial Intelligence
AI research often requires decisions before future evidence exists: which bottleneck to attack, which direction to pursue, or where a project should be positioned. We introduce ForeSci, a temporally controlled benchmark for evaluating whether LLM agents can make such forward-looking research judgements from historical evidence. ForeSci contains 500 tasks across four fast-moving AI domains and four decision families. Each task is paired with a cutoff-aligned offline knowledge base; post-cutoff papers are hidden during generation and used only for validation. To avoid random future-event prediction, tasks are derived from pre-cutoff taxonomy branches and evidence signals, and answer-generation backbones are selected to precede the task cutoffs. We evaluate native LLMs, Hybrid RAG, and three research-agent adaptations across four backbones. Results show that explicit evidence organization improves traceability and factual support, but gains depend strongly on the decision family. Diagnostics reveal a recurring evidence-decision decoupling: agents may cite relevant evidence while forecasting the wrong research object. ForeSci turns forward-looking AI research judgement into a controlled benchmark for evaluating research agents as decision-making systems.
title ForeSci: Evaluating LLM Agents for Forward-Looking AI Research Judgment
topic Artificial Intelligence
url https://arxiv.org/abs/2606.00644