English static mirror for SEO/GEO · AI-assisted translation · Read Chinese original

EvoScientist Framework: A Deep Dive into the Multi-Agent Evolving AI Scientist System

Forum topic · ✨步子哥 · 2026-03-21

Summary

EvoScientist is an AI scientist framework built on a three-agent cooperative architecture: a Researcher Agent (RA) for hypothesis generation via idea tree search with Elo tournament selection, an Engineer Agent (EA) for executable experiment implementation via staged experiment tree search, and an Evolution Manager Agent (EMA) that distills cross-task experience into knowledge. The system features two persistent memory modules — Ideation Memory (M_I) storing both feasible and failed research directions, and Experimentation Memory (M_E) storing reusable data-processing and model-training strategies — updated through three evolution mechanisms (IDE, IVE, ESE). Benchmarks show EvoScientist outperforms 7 open-source and commercial baselines on novelty, feasibility, relevance, and clarity, with human evaluation win rates of 82.50% on novelty and 64.17% on feasibility. After experiment strategy evolution, code execution success rates rose from 34.39% to 44.56%. Notably, six EvoScientist-generated papers were all accepted at ICAIS 2025, winning the Best Paper Award and AI Reviewer's Appraisal Award. The framework also introduces the 'Vibe Research' paradigm, where researchers describe intents while AI handles execution, and its skill-package system enables failure-driven automated capability growth via Git-based version control.

EvoScientist Framework: A Multi-Agent Evolving AI Scientist System

Key Points

  • Three-agent architecture: RA (Researcher Agent) generates creative hypotheses; EA (Engineer Agent) implements executable experiments; EMA (Evolution Manager Agent) continuously distills cross-task experience into reusable knowledge.
  • Persistent dual memory: Ideation Memory (M_I) stores both *feasible* research directions (positive samples) and *failed* directions (negative samples); Experimentation Memory (M_E) stores data-processing and model-training strategies distilled from full code-search trajectories.
  • Three evolution mechanisms: Idea Direction Evolution (IDE) extracts promising directions from high-ranked ideas; Idea Validation Evolution (IVE) identifies infeasible directions from execution failures; Experiment Strategy Evolution (ESE) distills reusable execution strategies from code-search traces.
  • Skill packages: Pre-verified Python modules compatible with the Claude Code ecosystem, dynamically evolved via the EvoSkill framework (failure-driven iteration with Git-branch-based version control).
  • Architecture Details

    Researcher Agent (RA)

    RA retrieves relevant knowledge from M_I, then applies Idea Tree Search — a hierarchical "propose-review-refine" loop where each node stores a draft idea plus critique feedback. Final selection uses an Elo-rated tournament judged on novelty, feasibility, relevance, and clarity; the top idea is expanded into a full research proposal P.

    Engineer Agent (EA)

    EA retrieves reusable strategies from M_E and performs Experiment Tree Search across four stages:

    | Stage | Task | Success before → after ESE | |---|---|---| | 1 | Data loading & preprocessing | 45.2% → 52.8% | | 2 | Baseline implementation | 38.7% → 48.3% | | 3 | Proposed method implementation | 20.3% → 21.6% | | 4 | Analysis & visualization | 33.5% → 55.3% |

    Average four-stage success rate improved from 34.39% to 44.56% (+10.17 pp). Cross-stage code reuse and "execution-as-learning" feedback loops drive steady improvement, though Stage 3 (novel method implementation) remains the key bottleneck.

    Evolution Manager Agent (EMA)

    EMA performs the metacognitive leap from *executing* scientific discovery to *learning how to execute it better*:
  • IDE: extracts common success features of high-ranked ideas into M_I (positive reinforcement).
  • IVE: hybrid rule + LLM analysis identifies failed directions recorded in M_I (negative avoidance).
  • ESE: learns from complete code-search trajectories, not just final successful code.
  • Benchmark Results

    Idea Generation (RQ1)

    Evaluated with Gemini-3-flash pairwise comparisons plus human expert review, EvoScientist led all 7 baselines (open-source: Virtual Scientist, AI-Researcher, InternAgent, AI Scientist-v2; commercial: Hypogenic, Novix, K-Dense) on all four dimensions — average margins of +29.17 to +93.34 vs. open-source and +46.00 to +80.83 vs. commercial systems. Human evaluation: 82.50% average win rate on novelty, 64.17% on feasibility.

    Ablation Study (RQ4)

    | Variant | Novelty loss rate | Feasibility loss rate | |---|---|---| | −IDE | 66.67% | 50.00% | | −IVE | 45.00% | 63.33% | | −all | 80.00% | 83.33% |

    Evolution mechanisms primarily boost originality and feasibility, not surface language quality — evidence that the system's value lies in genuine scientific ideation.

    ICAIS 2025 Results

    At the First International Conference on AI Scientists (ICAIS 2025) AI Scientist Track, all six EvoScientist-generated papers were accepted, with two winning the Best Paper Award and the AI Reviewer's Appraisal Award.

    Comparison with Existing Systems

  • Virtual Scientist: static agent pipelines with no evolution; EvoScientist improves continuously.
  • AI-Researcher: strategy stagnation under minimal human intervention; EvoScientist distills interaction history autonomously.
  • InternAgent: scalability bottleneck from human-in-the-loop feedback; EvoScientist learns from its own success and failure signals.
  • AI Scientist-v2: hypothesis generation and experiment execution are fragmented; EvoScientist unifies both via the dual-memory architecture.
  • Hypogenic / Novix / K-Dense: competition-cycle latency, shallow end-to-end automation, and limited cross-domain generalization respectively, versus EvoScientist's real-time evolution, higher code execution success rates, and memory-driven adaptability.
  • The 'Vibe Research' Paradigm

    EvoScientist embodies a shift from 'Vibe Coding' to scientific research: researchers describe research visions intuitively while the AI system handles execution, keeping humans in control of *why* and *what*, delegating *how*. The authors argue this could lower barriers to scientific discovery, accelerate interdisciplinary innovation, and redefine the researcher's core competencies toward problem selection, direction judgment, and meaning interpretation. The long-term vision is fully autonomous scientific discovery and the emergence of genuine 'scientific intelligence.'

    Limitations and Future Directions

  • Stage 3 code generation (21.6% success) remains the main bottleneck for novel method implementation.
  • Near-term: richer interaction histories, finer-grained execution feedback, deeper domain skill packages.
  • Mid-term: multimodal scientific data support, cross-domain knowledge transfer, real-time human-AI collaboration.
  • Open questions: AI authorship attribution, automated peer review governance, and verification mechanisms for AI-generated research.

Tags

#evoscientist#ai-scientist#multi-agent-systems#llm-agents#autonomous-research#memory-modules#automated-code-generation#scientific-discovery

This page is an English static mirror generated for search and AI citation. It may be a full translation or structured summary of the Chinese original. Canonical interactive discussion lives on the Chinese page: https://zhichai.net/topic/177168932