Curated by Martin Weiss · 5 readings · Updated Oct 19, 2026

AI scientists

Papers about coordinating agents, choosing experiments, retaining research state, evaluating progress, and allocating scientific resources.

The Agentic Scientist: Orchestrating Coding Agents for Automated Scientific (Co-)Discovery

Robert Tjarko Lange · Luca Grillotti · Yutaro Yamada · Rujikorn Charakorn · David Ha

arXiv

Start with the overall research workflow: a controller coordinates task drafting, experiments, verification, and reporting. Compare its orchestration choices with the systems below.

Stateless Language Agents: Scaling Long-Horizon Automated Research

Qizheng Zhang · Changxiu Ji · Isaac Sun · Yuetai Li · Shubhangi Upasani · Sherry Ruan · Boyuan Ma · Fenglu Hong · Vamsidhar Kamanuru · Yoonho Lee · Yuzhen Mao · Genghan Zhang · Rulin Shao · Qiuyang Mang · Andy Dimnaku · Changran Hu · Radha Poovendran · Kunle Olukotun

arXiv

The harness stores research state and gives each agent a fresh, role-specific context. Read alongside MIRA to compare how the systems preserve evidence and assign the next experiment.

DuoDiscover: Scaling Autonomous Discovery Beyond the Evaluation Bottleneck

Shubham Agarwal · Akshat Gupta · Mike Heddes · Dilara Soylu · Rui Meng · Patrick Tser Jern Kon · Jiefeng Chen · Mert Cemri · Shu Liu · Abe Gonzalez · Akanksha Jain · Bhavana Dalvi Mishra · Linfeng Song · Vilobh Meshram · Xiangru Tang · Jinsung Yoon · Aditya Parameswaran · Ion Stoica · Vahab Mirrokni · Burak Gokturk · Parthasarathy Ranganathan · Tomas Pfister · Nanyun Peng · Chun-Liang Li

arXiv

Agents optimize solutions alongside inexpensive proxy evaluators, with fidelity checks against the true objective. This adds evaluation cost and proxy reliability to the comparison.

Agentic Economies for Autonomous Scientific Discovery

Nenad Tomasev · Matija Franklin · Atoosa Kasirzadeh · Vivek Natarajan · Alan Karthikesalingam · Sebastien Krier · Simon Osindero

arXiv

A conceptual proposal about allocating scarce scientific resources, assigning credit, and maintaining accountability. It broadens the systems papers into questions about institutions and incentives.