ASI-Evolve: AI Accelerates AI

• URL: http://arxiv.org/abs/2603.29640v1
• Date: Tue, 31 Mar 2026 12:04:17 GMT
• Title: ASI-Evolve: AI Accelerates AI
• Abstract summary: ASI-Evolve is a framework for AI-for-AI research that closes the loop through a learn-design-experiment-analyze cycle.<n>To our knowledge, ASI-Evolve is the first unified framework to demonstrate AI-driven discovery across three central components of AI development.
• Score: 42.885465604426884
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Can AI accelerate the development of AI itself? While recent agentic systems have shown strong performance on well-scoped tasks with rapid feedback, it remains unclear whether they can tackle the costly, long-horizon, and weakly supervised research loops that drive real AI progress. We present ASI-Evolve, an agentic framework for AI-for-AI research that closes this loop through a learn-design-experiment-analyze cycle. ASI-Evolve augments standard evolutionary agents with two key components: a cognition base that injects accumulated human priors into each round of exploration, and a dedicated analyzer that distills complex experimental outcomes into reusable insights for future iterations. To our knowledge, ASI-Evolve is the first unified framework to demonstrate AI-driven discovery across three central components of AI development: data, architectures, and learning algorithms. In neural architecture design, it discovered 105 SOTA linear attention architectures, with the best discovered model surpassing DeltaNet by +0.97 points, nearly 3x the gain of recent human-designed improvements. In pretraining data curation, the evolved pipeline improves average benchmark performance by +3.96 points, with gains exceeding 18 points on MMLU. In reinforcement learning algorithm design, discovered algorithms outperform GRPO by up to +12.5 points on AMC32, +11.67 points on AIME24, and +5.04 points on OlympiadBench. We further provide initial evidence that this AI-for-AI paradigm can transfer beyond the AI stack through experiments in mathematics and biomedicine. Together, these results suggest that ASI-Evolve represents a promising step toward enabling AI to accelerate AI across the foundational stages of development, offering early evidence for the feasibility of closed-loop AI research.

Related papers

• ASI-Bench: At the Dawn of Artificial Superintelligence [46.937202634067766]
  ASI-Bench is the first benchmark to jointly evaluate AI systems' capabilities of innovative exploration and autonomous scientific execution.<n>It contains 60 project-level research tasks across 11 scientific domains and progressively reduces methodological guidance.<n>Across 18 state-of-the-art agent--model configurations, the average score drops from 50.91 with full methodological guidance to 29.10 with only the method specified.
  arXiv  Detail & Related papers  (2026-08-18T01:59:00Z)
• From AGI to ASI [65.04985565452489]
  Building human-level artificial general intelligence has moved from far-fetched speculation to being a concrete next-decade target for many of the largest AI organisations.<n>Report investigates how AI itself might continue to develop in a post-AGI world along the continuum of machine intelligence.
  arXiv  Detail & Related papers  (2026-06-10T21:14:10Z)
• AI Co-Scientist for Ranking: Discovering Novel Search Ranking Models alongside LLM-based AI Agents with Cloud Computing Access [60.37646376282544]
  We present an AI Co-Scientist framework that automates the full search ranking research pipeline.<n>This is the first study in the ranking community to utilize an AI Co-Scientist framework for algorithmic research.
  arXiv  Detail & Related papers  (2026-03-23T09:29:08Z)
• AI Phenomenology for Understanding Human-AI Experiences Across Eras [8.54750896635797]
  We argue for AI phenomenology: a research stance that asks "How did it feel?" when interacting with AI systems.<n>We motivate AI phenomenology as a framework that captures how alignment is experienced, negotiated, and updated between users and AI systems.
  arXiv  Detail & Related papers  (2026-03-09T23:26:46Z)
• AI Researchers' Views on Automating AI R&D and Intelligence Explosions [7.10975129848806]
  Many leading AI researchers expect AI development to exceed the transformative impact of all previous technological revolutions.<n>We interviewed 25 leading researchers from frontier AI labs and academia.
  arXiv  Detail & Related papers  (2026-02-13T23:24:05Z)
• AlphaGo Moment for Model Architecture Discovery [26.258185487878333]
  We present ASI-Arch, the first demonstration of Artificial Superintelligence for AI research (ASI4AI)<n>ASI-Arch is a fully autonomous system that shatters the constraint by enabling AI to conduct its own architectural innovation.<n>We conduct 1,773 autonomous experiments over 20,000 GPU hours, culminating in the discovery of 106 innovative, state-of-the-art (SOTA) linear attention architectures.
  arXiv  Detail & Related papers  (2025-07-24T03:57:27Z)
• Echoes of AI: Investigating the Downstream Effects of AI Assistants on Software Maintainability [5.677464428950146]
  This study investigates whether co-development with AI assistants affects software maintainability.<n> AI-assisted development in Phase 1 led to a modest speedup in subsequent evolution.<n>For habitual AI users, the mean speedup was 55.9%.
  arXiv  Detail & Related papers  (2025-07-01T14:24:37Z)
• AI Generations: From AI 1.0 to AI 4.0 [3.4440023363051266]
  This paper proposes that Artificial Intelligence (AI) progresses through several overlapping generations.<n>Each of these AI generations is driven by shifting priorities among algorithms, computing power, and data.<n>It explores the profound ethical, regulatory, and philosophical challenges that arise when artificial systems approach (or aspire to) human-like autonomy.
  arXiv  Detail & Related papers  (2025-02-16T23:19:44Z)
• Generative Diffusion-based Contract Design for Efficient AI Twins Migration in Vehicular Embodied AI Networks [55.15079732226397]
  Embodied AI is a rapidly advancing field that bridges the gap between cyberspace and physical space. In VEANET, embodied AI twins act as in-vehicle AI assistants to perform diverse tasks supporting autonomous driving.
  arXiv  Detail & Related papers  (2024-10-02T02:20:42Z)
• Large Language Model Empowered Next-Generation MIMO Networks: Fundamentals, Challenges, and Visions [80.7546823422894]
  Next-generation Multiple-Input Multiple-Output (MIMO) is expected to be intelligent and scalable.<n>We propose the concept of the generative AI agent, which is capable of generating tailored and specialized contents.<n>Case studies demonstrate the effectiveness of leveraging the generative AI agent for performance analysis.
  arXiv  Detail & Related papers  (2024-04-13T02:39:36Z)
• Exploration with Principles for Diverse AI Supervision [88.61687950039662]
  Training large transformers using next-token prediction has given rise to groundbreaking advancements in AI. While this generative AI approach has produced impressive results, it heavily leans on human supervision. This strong reliance on human oversight poses a significant hurdle to the advancement of AI innovation. We propose a novel paradigm termed Exploratory AI (EAI) aimed at autonomously generating high-quality training data.
  arXiv  Detail & Related papers  (2023-10-13T07:03:39Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.