AITOPAI热报
PapersSelected

ScienceBuddy: Recursive-in-Recursive Self-Improvement for Interactive Scientific Agents

arXiv cs.AI@Shuhan Xue et al. (13)
Paper
Read the original
Why we picked it

This research introduces a new paradigm for scientific AI that allows models to improve through interaction with researchers, which could be a significant development in the field.

We introduce and release ScienceBuddy, an interactive scientific research workspace that brings continually improving scientific agents into researchers' everyday workflows.

Body

ScienceBuddy supports researchers in carrying out scientific tasks while transforming their requests, feedback, and execution evidence into tasks and evaluation rubrics for continual learning.

At its core is recursive-in-recursive self-improvement, a paradigm that couples harness evolution with model reinforcement learning: the inner recursion improves the harness with the model fixed, while the outer recursion trains the model under the improved harness.

Harness evolution shapes training experience, and model learning creates new opportunities for harness adaptation.

We present case studies of researcher interaction, harness refinement, and model learning, with the benchmark cases spanning four scientific task families.

By releasing ScienceBuddy as a research product, we make this paradigm available to the scientific community and take a step toward discovery intelligence: scientific AI that advances through sustained collaboration with researchers and evolves alongside the research it supports.

Original · arXiv cs.AI

We introduce and release ScienceBuddy, an interactive scientific research workspace that brings continually improving scientific agents into researchers' everyday workflows. ScienceBuddy supports researchers in carrying out scientific tasks while transforming their requests, feedback, and execution evidence into tasks and evaluation rubrics for continual learning. At its core is recursive-in-recursive self-improvement, a paradigm that couples harness evolution with model reinforcement learning: the inner recursion improves the harness with the model fixed, while the outer recursion trains the model under the improved harness. Harness evolution shapes training experience, and model learning creates new opportunities for harness adaptation. We present case studies of researcher interaction, harness refinement, and model learning, with the benchmark cases spanning four scientific task families. By releasing ScienceBuddy as a research product, we make this paradigm available to the scientific community and take a step toward discovery intelligence: scientific AI that advances through sustained collaboration with researchers and evolves alongside the research it supports. Website: http://science-buddy.io

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include ScienceBuddy, recursive-in-recursive self-improvement, scientific agents, research workspace.

Source attribution: curated from arXiv cs.AI (@Shuhan Xue, Jianyuan Zhong, Ziyuan Nan, Wenbin Li, Zhaochen Yu, Jinchao Ding, Qiang Gao, Pengyu Zhan, Yuntong Zhang, Tian Cheng, Zhenfei Yin, Yingcheng Wu, Ling Yang).

#ScienceBuddy#recursive-in-recursive self-improvement#scientific agents#research workspace#continual learning

Keep reading

Recent stories with the same tag or section
ModelsBuildersSelected73°16:50
ScienceBuddy AI agent improvement

This approach solves the common problem of AI agent improvements being temporary, as fixes in chat typically don't transfer to new tasks. For developers building AI agents, this method provides a concrete framework to systematically improve performance by saving user corrections as test cases and iteratively upgrading both prompts and models. The research offers a practical solution for creating more reliable scientific AI assistants that can learn from interactions.

#ScienceBuddy#AI agents#prompt engineering#fine-tuning
rohanpaul_ai@rohanpaul_aiSource
PapersBuildersSelected04:26
Stanford Researchers Introduce Speculative Self-Distillation for LLM Fine-Tuning

The non-obvious insight is that you do not have to choose between on-policy accuracy and training cost — selective teacher intervention gets both, plus measurably less forgetting. Developers and teams doing domain adaptation or continued fine-tuning can use SSD-style distillation to add new knowledge without degrading existing capabilities at a fraction of the token budget. Anyone benchmarking against standard on-policy distillation pipelines should compare training time and retention metrics directly.

#Speculative Self-Distillation#NeurIPS 2026#fine-tuning#open-source models
Stanford AI Lab@StanfordAILabSource
PapersBuildersMulti-source04:16
Chinese contentMETR 在 Anthropic 的研究印证了"直觉测试"的有效性

METR 在 Anthropic 做的这项研究给 vibe-testing 背了书。如果你平时靠快速上手体验来判断模型好坏，这篇等于给你一个依据，还顺带聊了 Claude Sonnet 3.7 为什么是 2025 年最超预期的模型，做模型选型的人可以看看。

#Anthropic#METR#Claude Sonnet 3.7#模型评估
Event
Ethan Mollick@emollickSource
PapersBuildersSelected03:03
SteerDistill Quantizes LLM Agents via Closed-Loop Teacher Steering

The non-obvious point is that the teacher remains involved after deployment-style compression, correcting the student agent inside its execution loop instead of only at training time. Teams deploying agents on consumer GPUs or edge devices should evaluate whether this steering loop cuts their inference bill without breaking tool-use chains. Compared with static quantization pipelines like plain GPTQ-style post-training, the closed-loop approach targets the multi-step drift problem specific to agents.

#SteerDistill#quantization#agents
AK@_akhaliqSource
PapersBuildersOfficialSelected01:52
Apple Derives Theoretical Limits of KL-Regularized Language Model Alignment

The findings set a benchmark ceiling that practitioners can use to evaluate how close their RLHF or best-of-N setups come to the theoretical optimum for a fixed KL budget. Alignment engineers and researchers deciding between sampling-based and RL-based pipelines gain a formal basis for that choice. Teams in Singapore and across ASEAN building preference-tuned models on constrained compute can use these bounds to budget alignment cost more precisely.

#Apple#alignment#RLHF#reasoning models
Primary sourceApple ML ResearchSource