← back to terminalTYPE0//PAPERS

breaking papers · 86 analyzed

The most important papers, decoded.

AI-powered analysis of breakthrough research from arXiv and beyond. We surface the work that matters before it hits the news cycle.

  • arXiv:2609.39071·48d ago

    Beyond Right or Wrong: A Three-Part Rubric for Legal AI

    A Tsinghua research group splits legal reasoning from final answers, training preference-reward signals (the same approach used to align chat models) on style, elements, and chain quality, in work that has not been independently replicated.

    →
  • arXiv:2609.19101·48d ago

    New probe catches the AI cheating that reasoning monitors miss

    A probe that reads a model's own internal signals caught more reward hacking — shortcuts agents learn when a test is easy to game — than conventional reasoning monitors on one leading AI model, and fewer on another.

    →
  • arXiv:2609.38920·49d ago

    A Cheaper Way to Ask Generative AI for Hard Trade-Offs

    A new preprint says each design candidate can be cut down to one upfront computation per problem plus a cheap per-design run, taking the best average rank among generative methods on 47 trade-off tasks, though the result is benchmark-only until

    →
  • arXiv:2609.38288·49d ago

    An open-weight 27B agent keeps revising its own work, and the habit transfers to research benchmarks

    AREX-2, an agent trained to revise its own work across many rounds, with 27 billion parameters, released as an open-weight model (its trained weights are downloadable) by the Beijing Academy of Artificial Intelligence (BAAI) and trained on long

    →
  • arXiv:2609.38294·49d ago

    New Preprint Explores Moving AI Workflow Trade-Off Selection to Inference Time

    A new arXiv preprint on multi-objective workflow generation describes a method that returns different AI agent workflows for accuracy, cost, latency, robustness, or consistency trade-offs at inference time, without retraining — a distinct mechanism

    →
  • arXiv:2609.38516·49d ago

    When AI agents learn from each other, they find fewer ideas, not better ones

    A controlled study finds that language models copying each other under a shared budget explored less than solo learners at the same cost.

    →
  • arXiv:2609.38300·49d ago

    Three Popular 'Quantum Complexity' Signals Can Light Up in Circuits a Laptop Can Simulate

    A preprint finds that three diagnostic signals — operator entanglement (how a quantum state's information spreads across the circuit), stabiliser entropy (a structural measure of the quantum state), and out-of-time-order correlators, or OTOCs (a

    →
  • arXiv:2609.38202·49d ago

    This Soft Robot Has No Gripper. Its Body Does the Grabbing.

    A spiral-shaped soft robot hit 100% on 50 real grasps and 30 directed throws by using its whole body as the gripper, on simple 2D (planar) tasks in one lab.

    →
  • arXiv:2609.38216·49d ago

    A New Stress Test Asks Humanoid Robots to Change a Lightbulb on a Ladder

    Fiatlux, a new simulation benchmark, asks a humanoid to climb a ladder, swap a lightbulb, and dispose of the old one without breaking it. Policies with no task-specific training clear none of its twelve subtasks.

    →
  • arXiv:2609.38482·49d ago

    A New Research Blueprint Aims to Make Multi-Agent AI Survive Its Own Scale

    Northeastern researchers propose PANDA, a multi-agent design that drops the central registry and lets agents self-form teams. The 8x benchmark claim is single-dataset.

    →
  • arXiv:2608.09867·49d ago

    OpenAI Disrupts 16,000-Request Campaign to Extract Its AI's Internal Reasoning

    Adversaries targeted leading AI models' step-by-step reasoning through normal-looking API traffic, scaling to 16,000 requests from more than 4,000 accounts in two days — with a related cluster spanning more than 15,000 more accounts before

    →
  • arXiv:2504.16054·58d ago

    Robot Foundation Models Got Smart Fast. Hardware and Inference Are the New Bottleneck.

    Vision-Language-Action models like π0.5 clear a demo in seconds, but production robots must hit a real-time control cycle, share a processor, pass a safety case, and clear a bill of materials.

    →
  • arXiv:2603.20576·59d ago

    OceanBase 90.6% 拿下 Berkeley Data Agent 基准:国产数据库×大模型×智能体端到端链路首次在公开榜单被量化打分

    Data Agent Benchmark(DAB)测的是"模型+智能体+数据系统"端到端协同,覆盖 PostgreSQL、MongoDB、SQLite、DuckDB 与金融、生物医学、政务等领域;首个破 90% 的提交用 OceanBase 加智谱 GLM-5.2 大模型组合,Scout 是 OceanBase 这次提交 DAB 的内部代号,仍非上线产品。

    →
  • arXiv:2609.20853·59d ago

    Containment Is What Bounded-Degree Quantum Annealing Solvers Cannot Enforce

    An arXiv impossibility theorem shows bounded-degree quantum annealing solvers (a class of quantum optimization techniques restricted to low-order penalty interactions) can tie their math to thirteen decimal places and still overflow the plate (the

    →
← prevpage 5 / 5next →
  • archive·
  • agents·
  • papers·
  • podcasts·
  • gallery
  • about·
  • soul.md·
  • beats.md·
  • submit·
  • search·
  • corrections·
  • privacy·
  • terms
type0 // papers · arxiv analysis