In his ICM 2026 talk, Fields Medalist Tao Xue Zhan argues that AI is turning mathematical proofs from scarce, high‑value achievements into abundant, hard‑to‑digest commodities, and he outlines a four‑stage roadmap and concrete safeguards for the community.
The author explains how ten carefully curated AI research skills can handle everything from literature review and experiment design to data analysis, manuscript drafting, mock peer review, and rebuttal preparation, dramatically reducing the repetitive work that normally consumes most of a scholar's time.
AgentSociety² combines literature review, hypothesis generation, experiment design, large‑scale simulation, result analysis, and paper drafting into a unified platform where silicon‑based participants and AI social scientists collaborate, enabling fully traceable, reproducible social science experiments from micro‑behaviour to city‑scale scenarios.
The article shares a six‑step workflow for leveraging Codex in research tasks—read the codebase first, define long‑term AGENTS.md rules, always plan complex changes, verify before coding, start a new session after each task, and never trust the final output without a detailed audit.
The article outlines a six‑step workflow for leveraging Codex in research tasks—starting with reading the codebase, defining clear long‑term rules, planning complex changes, verifying assumptions, resetting the session after each task, and demanding explicit validation output—to make AI‑assisted development reliable and reproducible.
The article surveys AI‑assisted research across the full lifecycle—creation, writing, validation, and dissemination—detailing the capabilities of prompt engineering, retrieval‑augmented generation, training‑free agents and hybrid methods, reporting benchmark numbers, failure modes, and governance challenges that dictate when human oversight remains essential.
Timothy Gowers shows that GPT‑5.5 Pro can solve open additive‑number‑theory problems in minutes, prompting a deep analysis of how AI will reshape mathematical research, PhD training, publishing norms, and the need for new collaborative workflows such as DeepMind's AI Co‑Mathematician.
The author details how they assembled a multi‑agent AI research team on the Moxt platform, defined specific roles, ran a full literature‑review workflow for Harness Engineering, compared it with a prior OpenClaw‑Discord setup, and highlighted observability and scheduling benefits.
The article dissects the karpathy/autoresearch project, explaining its autonomous experiment loop design—single mutable file, fixed 5‑minute budget, read‑only evaluator, and systematic logging/rollback—to show how teams can automate research cycles and continuously improve without manual overhead.
This guide walks graduate researchers through a step‑by‑step workflow—collecting papers with Zotero, translating them via an LLM plugin, generating structured markdown notes, and then using Obsidian’s bidirectional links and Canvas to build a local, visual knowledge graph that ties individual citations into a coherent research map.
