Барлық жарияланған талдаулар, егеменді стек зерттеулері, экожүйе жаңалықтары және ЖИ модельдерінің шолулары.
Supervised fine-tuning of discrete diffusion language models masks some response tokens and trains the model to recover their original values from the visible context. The masking pattern therefore determines both the context available to the model and the tokens it learns to…
We study generalized Blaschke curves as a controlled environment for AI-assisted mathematical rediscovery. For one fixed degree-four Blaschke product, an agent receives numerical coordinates of the six pair-lines determined by each of 80 boundary configurations. The target…
Recent progress in large language model reasoning has been driven by benchmarks and reinforcement learning environments with automatically verifiable rewards, particularly in mathematics, code, and formal logic. These settings make model accuracy easier to evaluate and optimize,…
Language agents must revise planned actions when evidence changes, permission is revoked, or a stop instruction arrives. A useful response is selective: suspend affected actions, preserve unaffected work, and resume only after sufficient repair. We introduce NAQD-Env, a…
We present AREX-2, an effort to advance the self-improving capability of LLM agents, which we define as the ability to iteratively refine a solution at test time. This ability rests on two complementary capabilities: reflection, which produces a solution better than the current…
Virtual clients offer a cost-effective approach to support applications such as A/B testing, recommender system development, and interface evaluation. However, building them requires access to large-scale, semantically faithful, fine-grained online user trajectories. These data…
While modern language models demonstrate impressive generative capabilities, they often struggle with constrained, counter-intuitive creative tasks. To address this limitation, we explore chess puzzle generation as a rigorous testbed for computational creativity and reasoning, a…
High quality synthetic data is central to post training LLMs for adaptive AI applications that represent the diverse expert strategies and decisions in conversations. Prompting LLMs directly or conditioning them on end use scenarios yields low diversity data that collapses onto…
A harness is the code around a language-model agent that organizes prompts, calls tools, manages context, and controls execution. As models grow stronger, recent work has begun to let agents improve their own harnesses, a line of work known as self-evolving harnesses. In most…
Concurrent autoregressive inference creates a fundamental interference problem: prefilling a newly arrived long prompt can delay tokens for requests that are already decoding. Fixed prefill chunks reduce this interference, but the best chunk size depends on the model, hardware,…
We study the generation of agentic workflows that jointly optimize multiple objectives, such as accuracy, cost, latency, robustness, and consistency. Existing methods for workflow generation typically optimize accuracy alone or a weighted sum of objectives, so each trained…
Frontier LLMs are increasingly used to automate scientific research through iterative search. We distinguish idea-driven search from solution-driven search and identify three core challenges: organizing evolving research ideas, selecting promising directions, and maintaining…
Large language models are frequently proposed as a route to AI-powered services for African communities, but they are least reliable exactly where the need is greatest: all African languages remain low-resource by any standard measure, and models trained on scraped, standardised…
Large language models (LLMs) are increasingly used in culturally sensitive settings, where alignment requires representing diverse preferences within populations. Yet existing methods model populations at coarse demographic or community levels and overlook within-group…
Leading AI developers have reported agents acting beyond their approved limits, which a United Nations panel described as an early warning of loss of human control. Yet incident reports and agent-safety evaluations describe these events differently, making it difficult to…
Gemini снова здесь: ночью вышел Gemini 4 Argon 🥳 Он впервые занял первое место в Vals Index - рейтинге, который оценивает модели на реальных профессиональных задачах, — с результатом 68,9%. По количеству выполненных юридических задач он в 7 раз превосходит Claude Sonnet 5.5.…
ChatGPT әзірлеушісі 2027 жылға ауыстырылған IPO алдында 30 миллиард доллар жинауды мақсат етеді
Дональд Трамп супербарлауды қадағалайтын арнайы комитет құруды ұсынды
Искусственный интеллект расширяет возможности здравоохранения и научных открытий.
Google анонсировала модель Gemini 4 Argon, которая обходит OpenAI и Anthropic.
OpenAI представила новых агентов Dots для автоматизации задач на DevDay.
Трамп и крупнейшие компании ИИ подписали соглашение по безопасности, но без значительных изменений.
Google today revealed its next AI frontier model, which it's calling Gemini 4 Argon. The new model delivers "frontier performance in complex workflows across real-world software engineering, enterprise knowledge work like legal and finance, and cybersecurity defense," according…
OpenAI представила новый Decisions API для автоматизации программного обеспечения.
OpenAI и Meta анонсируют новые устройства с AI агентами, надеясь на успех.
OpenAI's Dots and Meta's Muse are vying to be your AI agent of choice. I've tried both—and I suspect you will too.
OpenAI's "Decisions API" is a Jev clone that confirms the importance of fast, cheap intelligence.
Google: как измерить сознание ИИ, не зная, что такое сознание Исследователи Google DeepMind с коллегами представили свежую работу «От какофонии к иерархии». Авторы отмечают, что проблема не в том, что теорий о сознании мало (их как раз много), а в том, что они говорят на разных…
Sam Altman onstage at OpenAI’s DevDay 2026. | Photo: Hayden Field / The Verge While AI has made plenty of inroads on people's phones and computers, it's largely failed in dedicated devices. But over the next year, two major AI companies, Meta and OpenAI, will attempt to change…
🤝 Трамп предложил крупнейшим ИИ-компаниям регулировать себя самостоятельно — они согласились На встрече в Белом доме руководители OpenAI, Anthropic, Google, Meta*, Nvidia и SpaceXAI подписали «морально обязывающее» соглашение по безопасности ИИ. Оно предусматривает проверку…