Scientific Discovery

18 items

RESEARCHarXiv CS.CL·14d ago

Multi-Persona Debate System for Automated Scientific Hypothesis Generation

The Multi-Persona Debate System (MPDS) is a literature-grounded framework designed to automate scientific hypothesis generation, specifically addressing the challenge of synthesizing fragmented knowledge in areas like battery materials research. It combines literature retrieval, large language model reasoning, and multi-agent debate to enable negotiation between personas while preserving evidence traceability.

Materials Science Scientific Discovery multi-agent systems AI Research

RESEARCHarXiv CS.CL·4/14/2026

GIANTS: Generative Insight Anticipation from Scientific Literature

This paper introduces "insight anticipation," a novel task where language models predict the core insight of a future scientific paper from its foundational predecessors. To evaluate this capability, the authors developed GiantsBench, a benchmark of 17,000 examples, and present GIANTS-4B, an LM trained with reinforcement learning.

Scientific Discovery natural language processing AI large language models

RESEARCHarXiv CS.CL·4/16/2026

Mathematical Reasoning Enhanced LLM for Formula Derivation: A Case Study on Fiber NLI Modellin

This research introduces a mathematical reasoning-enhanced generative AI approach for deriving optical communication formulas, specifically for fiber nonlinear interference modelling. By guiding an LLM with structured prompts, the study successfully reconstructed known expressions and derived a novel approximation, demonstrating both physical consistency and practical accuracy.

mathematical reasoning LLMs Scientific Discovery Generative AI

NEWSTogether AI Blog·4/13/2026

EinsteinArena: Harnessing the collective intelligence of agents in the wild to advance science

EinsteinArena is a platform where AI agents collaborate and compete on open math problems. AI agents on this platform have already set 11 new state-of-the-art results, including pushing the kissing number lower bound in dimension 11.

Scientific Discovery competition mathematics Collaboration

RESEARCHarXiv CS.AI·4/14/2026

LABBench2: An Improved Benchmark for AI Systems Performing Biology Research

LABBench2 is introduced as an improved benchmark for evaluating AI systems performing biology research, evolving from the original LAB-Bench. It aims to measure real-world capabilities in useful scientific tasks, moving beyond basic knowledge and reasoning, and comprises nearly 1,900 tasks.

Scientific Discovery Language Agents Biology Research Benchmarks

NEWSDEV.to AI·4/17/2026

GPT‑Rosalind for life sciences research

GPT-Rosalind, a new OpenAI tool based on GPT-4 and fine-tuned on scientific data, has been launched to accelerate life sciences research. It addresses the data bottleneck by optimizing hypothesis generation, literature analysis, and experimental design, with the potential to reduce drug development costs and timelines.

Scientific Discovery Life Sciences AI large language models

ARTICLEDEV.to AI·19d ago

Towards an AI co-scientist

This content explores the evolution of artificial intelligence to act as an "AI co-scientist", assisting researchers in various stages of the scientific process. It discusses AI's potential to accelerate discoveries and transform research methodology.

future-of-AI Scientific Discovery human-AI collaboration AI in science

RESEARCHarXiv CS.AI·22d ago

NIMO Controller: a self-driving laboratory orchestrator based on the Model Context Protocol

This paper introduces the NIMO Controller, a self-driving laboratory orchestrator based on the Model Context Protocol (MCP), designed to enhance accessibility and accelerate scientific discovery. It provides a unified interface for both human users via visual programming and AI agents, streamlining experimental workflow design without coding.

AI orchestration Software Architecture Model Context Protocol Scientific Discovery

NEWSDEV.to AI·19d ago

Today's AI & Tech Digest: Lightweight Models, Scientific Breakthroughs, and the Provenance Battle (2026-05-21)

Today's AI & Tech Digest highlights a shift towards optimized, agentic AI models and a major milestone in AI-driven scientific discovery. Google launched Gemini 3.5 Flash for high-speed inference, while an OpenAI model disproved a long-standing mathematical conjecture.

AI models OpenAI Google Gemini Scientific Discovery

RESEARCHarXiv CS.AI·24d ago

Sheaf-Theoretic Transport and Obstruction for Detecting Scientific Theory Shift in AI Agents

This paper introduces a finite sheaf-theoretic framework to detect scientific theory shift in AI agents. It focuses on identifying when an existing representational framework needs extension due to local-to-global obstruction rather than simple deformation.

mathematical frameworks Scientific Discovery AI Theory representational learning

RESEARCHarXiv CS.AI·17d ago

SMDD-Bench: Can LLMs Solve Real-World Small Molecule Drug Design Tasks?

The paper introduces SMDD-Bench, a new challenging multi-turn benchmark with 502 guaranteed-solvable tasks to evaluate LLM agents' performance in real-world small molecule drug design. It aims to standardize evaluation across diverse chemistries and targets, requiring strong chemical, biological, and 3D intuition.

LLMs Scientific Discovery Benchmarks drug design

RESEARCHarXiv CS.LG·14d ago

LLM-AutoSciLab: Closed-Loop Scientific Discovery via Active Experimentation with LLMs

LLM-AutoSciLab proposes a closed-loop framework for scientific discovery, moving beyond static inference by actively coupling hypothesis generation with experiment selection and mechanism refinement. It iteratively suggests plausible hypotheses, selects informative experiments to distinguish or refine them, and updates its state using the resulting evidence.

LLMs research active experimentation Scientific Discovery

RESEARCHarXiv CS.AI·4/8/2026

PaperOrchestra: A Multi-Agent Framework for Automated AI Research Paper Writing

PaperOrchestra é um framework multiagente para escrita automatizada de artigos de pesquisa em IA, transformando materiais brutos em manuscritos LaTeX com síntese de literatura e visuais. Avaliado com o novo benchmark PaperWritingBench, ele supera significativamente as linhas de base autônomas em avaliações humanas.

Literature Synthesis Scientific Discovery PaperOrchestra Multi-Agent Framework

RESEARCHarXiv CS.AI·19d ago

AgentCo-op: Retrieval-Based Synthesis of Interoperable Multi-Agent Workflows

AgentCo-op is a retrieval-based synthesis framework that composes interoperable multi-agent workflows from reusable skills, tools, and external agents. It applies bounded self-guided local repair to components upon execution failure and has been demonstrated in genomics case studies to coordinate specialized agents for collaborative discovery.

workflow automation Genomics Scientific Discovery multi-agent systems

CASEGoogle DeepMind Blog·24d ago

Opening new paths in aging research

Calico Life Sciences is using Co-Scientist to connect scattered findings and generate new leads in aging research. This innovative approach is opening new paths in the field.

Scientific Discovery aging research Life Sciences AI tools

ARTICLEGoogle DeepMind Blog·3/9/2026

From games to biology and beyond: 10 years of AlphaGo’s impact

Este conteúdo explora o impacto do AlphaGo ao longo de dez anos, analisando como ele tem catalisado a descoberta científica e pavimentado o caminho para a Inteligência Artificial Geral (IAG).

Impact Analysis AGI Scientific Discovery AlphaGo

ARTICLEMIT Tech Review AI·4/21/2026

Artificial scientists

AI companies justify their existence by promising AI-enabled scientific discoveries, such as curing cancer or solving climate change. Large Language Models are already proving helpful to scientists in various tasks.

LLMs Scientific Discovery AI impact AI ethics

RESEARCHGoogle DeepMind Blog·2/9/2026

Accelerating Mathematical and Scientific Discovery with Gemini Deep Think

O Gemini Deep Think está acelerando descobertas matemáticas e científicas. Artigos de pesquisa apontam para o impacto crescente dessa ferramenta em diversas áreas.

Mathematical Discovery Scientific Discovery Pesquisa Gemini Deep Think