# Scientific Discovery

Large language models can now construct and keep revising scientific principles during a discovery run, rather than leaving that translation to human experts. Doing it by hand needs domain expertise, time and effort, and bottlenecks discovery. Principia compares related inputs and their observed outcomes to write causal hypotheses, clusters similar inputs to induce recurring principles, then revises them with new observations. Across molecular optimization, protein engineering, gene perturbation and molecular property prediction, it beat baselines using internal knowledge, experience summaries or web retrieval; on five molecular optimization tasks it averaged AUC@10 0.279 versus 0.250 for the experience-summary baseline.

Large language models can now construct and keep revising scientific principles during a discovery run, rather than leaving that translation to human experts. Doing it by hand needs domain expertise, time and effort, and bottlenecks discovery. Principia compares related inputs and their observed outcomes to write causal hypotheses, clusters similar inputs to induce recurring principles, then revises them with new observations. Across molecular optimization, protein engineering, gene perturbation and molecular property prediction, it beat baselines using internal knowledge, experience summaries or web retrieval; on five molecular optimization tasks it averaged AUC@10 0.279 versus 0.250 for the experience-summary baseline.

A new benchmark turns senior researchers' corrections of scientific AI agents into 248 two-way judgment questions: given only the project record from before the correction, four of five frontier models more often chose the original plan that a human had overruled, topping out at 63.5% against 73.4% for fresh researchers. The cases come from 26 real projects across six fields run over roughly half a year, and each asks a model to choose between the original plan and the human-corrected alternative.

A new benchmark turns senior researchers' corrections of scientific AI agents into 248 two-way judgment questions: given only the project record from before the correction, four of five frontier models more often chose the original plan that a human had overruled, topping out at 63.5% against 73.4% for fresh researchers. The cases come from 26 real projects across six fields run over roughly half a year, and each asks a model to choose between the original plan and the human-corrected alternative.

LANTERN trains a linear classifier on Qwen3-32B activations to rank 50 million candidate relations among the 10,000 most-referenced OEIS sequences, then applies staged filtering, hypothesis generation, executable verification and analytical checking to produce 62 verified relations without an existing OEIS cross-reference; a content screen retained 13, nine of them informative or insightful, four not found in the OEIS or a targeted literature search, with the end-to-end process taking under 8 hours.

LANTERN trains a linear classifier on Qwen3-32B activations to rank 50 million candidate relations among the 10,000 most-referenced OEIS sequences, then applies staged filtering, hypothesis generation, executable verification and analytical checking to produce 62 verified relations without an existing OEIS cross-reference; a content screen retained 13, nine of them informative or insightful, four not found in the OEIS or a targeted literature search, with the end-to-end process taking under 8 hours.

The work introduces the Agentic Idea Manager (AIM), a fully autonomous framework that makes research ideas explicit search objects: a Bayesian-optimization-inspired Agentic Surrogate organizes ideas into semantic clusters and produces ordinal promisingness estimates, an Agentic Acquisition mechanism balances exploration and exploitation at cluster and idea level, a Solution Auditor checks idea-solution integrity, and a Resource Planner adaptively allocates parallel branches under a fixed budget; on 10 AutoLab tasks AIM reaches 67.0% average on System Optimization and 55.8% on long-horizon Model Development & CUDA, exceeding the strongest baseline ScientistOne by 1.6 and 4.9 percentage points and reaching that baseline's best score up to 3.1x faster in wall-clock time.

The work introduces the Agentic Idea Manager (AIM), a fully autonomous framework that makes research ideas explicit search objects: a Bayesian-optimization-inspired Agentic Surrogate organizes ideas into semantic clusters and produces ordinal promisingness estimates, an Agentic Acquisition mechanism balances exploration and exploitation at cluster and idea level, a Solution Auditor checks idea-solution integrity, and a Resource Planner adaptively allocates parallel branches under a fixed budget; on 10 AutoLab tasks AIM reaches 67.0% average on System Optimization and 55.8% on long-horizon Model Development & CUDA, exceeding the strongest baseline ScientistOne by 1.6 and 4.9 percentage points and reaching that baseline's best score up to 3.1x faster in wall-clock time.

The work introduces EvoDuet, a bi-level method that co-evolves solutions and web search queries with fixed model parameters: at each iteration a knowledge-gap-based retrieval gate decides whether to retrieve new documents, reuse stored ones, or proceed without them, an inner loop refines queries and ranks documents by the solution scores they are predicted to yield, and an outer loop generates candidates in parallel and records evaluated outcomes for later searches; across 21 optimization tasks with one candidate per iteration it raises OpenEvolve's normalized discovery gain from 74.1% to 78.0% with GPT-5.6-Luna and from 61.3% to 82.3% with Gemini-3.8-Flash, Qwen3.

The work introduces EvoDuet, a bi-level method that co-evolves solutions and web search queries with fixed model parameters: at each iteration a knowledge-gap-based retrieval gate decides whether to retrieve new documents, reuse stored ones, or proceed without them, an inner loop refines queries and ranks documents by the solution scores they are predicted to yield, and an outer loop generates candidates in parallel and records evaluated outcomes for later searches; across 21 optimization tasks with one candidate per iteration it raises OpenEvolve's normalized discovery gain from 74.1% to 78.0% with GPT-5.6-Luna and from 61.3% to 82.3% with Gemini-3.8-Flash, Qwen3.