#AlphaFold

[hellsite-proteins]

Hi!

the goal of this blog is to use the strings of letters from text posts matching proteinogenic amino acids into 3D structures using AlphaFold. i am not affiliated with AlphaFold in any way, just using it to create and share some neat images. the idea for this was heavily inspired by @hellsitegenetics .

to make these, i will be removing any letters not matching proteinogenic amino acids, as well as any other characters, and plugging the resulting sequence into AlphaFold3. while i tried in some early posts, unusual amino acids unfortunately do not work in my structures. therefore, B, J, O, U, X and Z will all be omitted. in the final image i will usually only be showing the cartoon and hide the surface and side chains, for the sake of making a somewhat more pleasant image. i now use ChimeraX to open and show structures. they may not always be accurate depending on the string of letters, so unless i say anything else, assume that none of it is reliable!

please send me any posts you want to see made into structures :) i can't promise i'll answer fast but i will do my best

some of the more popular protein memes:

never gonna give you up, leitner rant, bee movie intro, spiders georg, ihnmaims hate, man door hand hook car door

and of course: the full bee movie script

here are some tags i tend to use to categorize things:

#protein asks for inbox submissions

#protein memes and #protein songs for memes and songs

#protein info if i'm answering science questions or explaining things

[drakefruitddg]

it's definitely widely misused but I think it's an overstep to say we never needed generative AI

It's miles better than Google translate ever has been (in the languages that model has been trained on at least)

There are a lot of scientific applications to the underlying technology behind these models and the research done on how to make them better is insanely helpful for furthering this stuff, a (kind of overused) example being deepmind's alpha fold, which solved a problem in the medical field that's been previously impossible

I know it can be easy to see grifters and tech bros being annoying as usual, but you can make fun of them without throwing the baby out with the bathwater if that makes sense.

I think a lot of what pro-AI people are really wanting is stuff that already exists but they don't know it's out there like

can't format a work email? templates

don't know how to write a resume? templates

writing a thank you card or a condolences card or a wedding invitation? templates templates templates

not sure how to format your citations in MLA or whatever format? citationmachine.net

summary of something you're reading for school/work? cliffsnotes.com

recipe based on ingredients in your fridge? whatsintherefrigerator.com

there's a million more like, guys, we don't need AI, we never needed generative AI

[er-10-media]

Лекарство от рака, созданное ИИ, готовится к испытаниям

New Post has been published on https://er10.kz/read/it-novosti/lekarstvo-ot-raka-sozdannoe-ii-gotovitsja-k-ispytanijam/

Лекарство от рака, созданное ИИ, готовится к испытаниям

Лекарство от рака, разработанное компанией Isomorphic Labs (Alphabet) с помощью ИИ-модели DeepMind AlphaFold 3, готовится к клиническим испытаниям на людях, пишет Gizmodo.com

Компания Isomorphic Labs приближается к важной вехе в мире разработки лечебных препаратов. При помощи ИИ-модели DeepMind AlphaFold 3 она создала перспективное лекарство от рака.

Более того, с помощью этой ИИ-платформы компания планирует создать универсальное средство для поиска препаратов от всех болезней «одним кликом», ускорить процесс их открытия, снизить затраты и повысить шансы на успех в лечении недугов.

ИИ-технология AlphaFold прославилась предсказанием трехмерной структуры белков. Эта способность дала ученым более ясное представление о том, как белки взаимодействуют с другими молекулами, что крайне важно для разработки новых лекарств.

Isomorphic Labs взяла эту основу и создала команду, которая объединяет исследователей ИИ с экспертами фармацевтической отрасли. Работа компании сосредоточена на излечении самых опасных заболеваний. Isomorphic Labs заклю��ила партнерские соглашения с крупными фармацевтическими фирмами, включая Novartis и Eli Lilly.

Эти сотрудничества направлены на поддержку как существующих ��рограмм разработки лекарств, так и на развитие новых собственных разработок.

Ранее сообщалось, что ИИ-модель AlphaFold предсказывает взаимодействие молекул жизни.

[jcmarchi]

TheSequence Radar #674: Transformers in the Genome: How AlphaGenome Reimagines AI-Driven Genomics

New Post has been published on https://thedigitalinsider.com/thesequence-radar-674-transformers-in-the-genome-how-alphagenome-reimagines-ai-driven-genomics/

TheSequence Radar #674: Transformers in the Genome: How AlphaGenome Reimagines AI-Driven Genomics

A model that could advance the future genomics.

Created Using GPT-4o

Next Week in The Sequence:

Knowledge: An intro to the world of multi-agent benchmarks.

Engineering: Let’s hack with the Gemini CLI.

Opinion: Why circuits could be the answer to AI interpretability?

Research: AlphaGenome deep dive.

Let’s Go! You can subscribe to The Sequence below:

TheSequence is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

📝 Editorial: Transformers in the Genome: How AlphaGenome Reimagines AI-Driven Genomics

I have been obsessed with AI in genetics for some time so I couldn’t write about anything else today other than DeepMind’s new model: AlphaGenome!

AlphaGenome merges some of the best -established techiques in AI-driven genomics such as large-scale sequence context with base-pair precision to chart the regulatory genome in a way never before possible. The model’s four-headed architecture digests up to one million contiguous base pairs in a single pass, outputting synchronized predictions for chromatin accessibility, transcription-factor occupancy, RNA expression, splicing, and 3D genome architecture. This unified approach replaces fragmented, single-modality pipelines—each requiring separate models and datasets—with one cohesive model that excels across tasks, streamlining variant effect analysis for researchers.

At its core, AlphaGenome marries convolutional layers, which capture local nucleotide motifs analogous to transcription-factor binding sites, with transformer modules that integrate distal regulatory elements hundreds of kilobases apart. DeepMind’s design eschews downsampling, ensuring every nucleotide contributes to high-resolution inferences. As functional genomics datasets from consortia like ENCODE, GTEx, and 4D Nucleome expand, this backbone stands ready to unveil regulatory grammar buried deep in non-coding DNA.

Traditional genomics models often excel at one signal—SpliceAI for splicing, ChromBPNet for chromatin state—necessitating an ensemble of tools to profile variant consequences fully. AlphaGenome’s simultaneous four-headed predictions eliminate this bottleneck, revealing cross-modal interactions—e.g., how a variant that disrupts a splice site may also alter local chromatin loops—opening novel avenues for mechanistic insight.

In benchmark evaluations spanning 24 sequence-prediction and 26 variant-effect tasks, AlphaGenome matches or surpasses specialized baselines in over 90% of cases. It outperforms SpliceAI, ChromBPNet, and other state-of-the-art models by significant margins, all while completing variant-effect scans in under a second—transforming in silico hypothesis testing from minutes or hours to real-time speed.

The genomics market in 2025 stands at an inflection point: cloud-based sequencing costs have halved over five years, single-cell and long-read technologies have become routine, and multi-omic datasets proliferate. Yet, analytical bottlenecks limit the translation of raw data into actionable insights. AlphaGenome arrives precisely when biotechnology and pharmaceutical companies demand scalable, AI-driven interpretation to bridge the gap from variant discovery to biological understanding. Its ability to standardize and accelerate regulatory variant annotation is poised to catalyze next-generation diagnostic tools, precision therapeutics, and synthetic biology platforms, redefining competitive advantage in a data-saturated market.

DeepMind’s preview API grants non-commercial researchers early access to AlphaGenome, democratizing large-scale regulatory predictions. From pinpointing causal non-coding mutations in disease cohorts to engineering synthetic enhancers with bespoke cell-type specificity, this open sandbox invites collaborative breakthroughs across academia and industry.

If AlphaFold decoded protein structures, AlphaGenome now deciphers the regulatory code—the “dark matter” governing gene expression. As single-cell, long-read, and cross-species datasets proliferate, the model’s extensible architecture promises seamless integration of new modalities. The future of genomics is computational, and AlphaGenome lights the path forward: an intellectual and technological leap toward understanding—and ultimately rewriting—the language of life.

🔎 AI Research

AlphaGenome: advancing regulatory variant effect prediction with a unified DNA sequence model

AI Lab: Google DeepMind Summary: AlphaGenome is a deep learning–based sequence-to-function model that ingests one megabase of DNA sequence and predicts thousands of functional genomic tracks—including gene expression, transcription initiation, chromatin accessibility, histone modifications, transcription factor binding, chromatin contact maps, and splicing—at single-base-pair resolution. Trained on both human and mouse experimental data, it unifies long-range sequence context with high prediction resolution, outperforming prior methods and enabling comprehensive in silico characterization of regulatory variant effects.

Confidential Inference Systems: Design Principles and Security Risks

AI Lab: Pattern Labs / Anthropic Summary: This whitepaper defines the architecture of a “confidential inference system” that leverages hardware-based Trusted Execution Environments (TEEs) to protect both user data (model inputs/outputs) and model assets (weights and architecture) during AI inference workloads. It further details reference designs for secure model provisioning, enclave build environments, service provider guarantees, and a comprehensive threat model to mitigate systemic and implementation-introduced risks.

USAD: Universal Speech and Audio Representation via Distillation

AI Lab: MIT CSAIL Summary: USAD distills knowledge from multiple domain-specific self-supervised audio models into a single student network capable of representing speech, music, and environmental sounds. By training on a diverse multimedia corpus with layer-to-layer distillation, it achieves near state-of-the-art performance across frame-level speech tasks, audio tagging, and sound classification.

UniVLA: Unified Vision-Language-Action Model

AI Lab: CASIA / BAAI / Tsinghua University / HKISI Summary: UniVLA reformulates vision, language, and robotic actions into shared discrete tokens and learns them jointly in an autoregressive transformer, eliminating separate modality encoders or mapping modules. This unified approach, trained on large-scale video datasets, sets new benchmarks on multi-stage robot manipulation tasks like CALVIN and LIBERO.

ProtoReasoning: Prototypes as the Foundation for Generalizable Reasoning in LLMs

AI Lab: ByteDance Seed / Shanghai Jiao Tong University Summary: ProtoReasoning introduces “reasoning prototypes”—abstract Prolog and PDDL templates—that capture common logical patterns across diverse tasks and guides LLMs to translate problems into these prototypes. Automated prototype construction and verification via interpreters boosts model generalization and reasoning performance on out-of-distribution benchmarks.

Reinforcement Learning Teachers of Test-Time Scaling

AI Lab: Sakana AI Summary: This work trains compact “Reinforcement-Learned Teachers” that ingest both questions and ground-truth solutions to learn dense rewards aligned with student performance, departing from sparse-reward paradigms. A 7B-parameter teacher model surpasses much larger reasoning models on competition-level math and science benchmarks and transfers zero-shot to novel tasks.

🤖 AI Tech Releases

Gemma 3n

Google released a full version of Gemma 3n, its mobile optimized model.

Gemini CLI

Google open sourced Gemini CLI, a coding terminal agent powered by Gemini.

Manus Browser

Manus released an agentic browser.

Qwen-VLo

Alibaba open sourced Qwen-VLo, an image understanding and generation model.

🛠 AI in Production

Project Vend

Anthropic showcased Project Vend, a system that allows Claude to run a small shop.

Ray at Pinterest

Pinterest shares how they scale end-to-end ML pipelines with Ray.

📡AI Radar

Meta has successfully recruited Lucas Beyer, Alexander Kolesnikov, and Xiaohua Zhai—founders of OpenAI’s Zurich office—to its new “superintelligence team” in what’s being called Zuckerberg’s latest recruiting victory.

Anthropic launched its Economic Futures Program to support research and policy development.

Uber is in talks with Travis Kalanick to find autonomous car company Pony.AI.

Prediction market Kalshi closed a $185 million Series B round led by Paradigm at a $2 billion post-money valuation, even as rival Polymarket reportedly eyes a $200 million raise.

Data management firm Rubrik announced an agreement to acquire Predibase to speed enterprise adoption of agentic AI—from pilot deployments to production at scale.

Battery startup Nascent Materials emerged from stealth after raising $2.3 million to commercialize an energy-efficient process that produces uniformly sized LFP cathode particles for higher-density, lower-cost batteries.

E-commerce veteran Julie Bornstein’s startup Daydream is launching an AI-powered chatbot tailored for fashion shopping following its $50 million seed round.

AI medical scribe Abridge secured $300 million in a Series E to double its valuation to $5.3 billion, led by Andreessen Horowitz with participation from Khosla Ventures.

Voice-to-text app Wispr Flow raised $30 million in Series A funding from Menlo Ventures (with NEA, 8VC, and angel investors) to scale its AI-powered dictation software across Mac, Windows, and iOS.

Andy Konwinski, co-founder of Databricks and Perplexity, pledged $100 million of his own funds via the Laude Institute to back AI research grants and the new AI Systems Lab at UC Berkeley.

Legal-focused AI startup Harvey AI raised $300 million in Series E funding at a $5 billion valuation—just four months after its prior $3 billion round—to expand its automation tools beyond law into professional services.

European challenger bank Finom closed a €115 million Series C led by AVP, bringing its total funding to ~$346 million as it ramps up AI-enabled accounting and targets 1 million SMB customers by 2026.

Creatio unveiled its 8.3 “Twin” release, embedding a unified conversational interface and new role-based AI agents for CRM and workflow automation along with AI-powered no-code development tools at no extra cost.

Nvidia shares have surged back to a record week, positioning the company within striking distance of a $4 trillion market capitalization as demand for its AI chips continues to accelerate.

Audos, the AI-powered startup studio, aims to democratize entrepreneurship by using AI agents and social-media distribution to launch up to 100,000 companies annually without taking equity.

[durutanhaber]

Yapay Zekâyla Şekil Alan Dünya: Proteinlerden Finans Katlarına

Bir proteinin nasıl göründüğü sadece bir bilimsel merak değil, insanlığın kaderini etkileyebilecek kadar kritik bir bilgi. Çünkü bu şekil, hastalıkları anlamaktan yeni ilaçlar geliştirmeye kadar birçok alanda kilit rol oynuyor. Ancak bir proteinin üç boyutlu yapısını çözmek, yıllarca süren zorlu bir çaba gerektiriyor. Bugün dünya üzerinde bilinen yaklaşık 200 milyon protein var. Ama bilim…

[peetum]

I may kid around on tumblr a lot, but I have a bachelor's in Biology and a Doctorate of Pharmacy. Protein folding was some of the stuff that would most break students brains. Even just the phrase "now flip the chair" for simple basic ring structures gives me a mini moment. THIS is the kind of thing AI should be used for - analytics and quantitative- Not generative (scraping artwork off the web and making an amalgamation). It's something that could potentially solve sooo many problems related to health and healthcare for folks.

(Source)

[ailifehacks]

How AI in Protein Folding Is Revolutionizing Structural Biology and Drug Discovery

Explore how AI in Protein Folding is transforming structural biology, drug discovery, and protein structure prediction across the USA and globally. What Is Protein Folding and Why AI Matters in Structural Biology Proteins perform critical functions in the human body, but their functions depend entirely on their unique 3D structure. Predicting how a linear amino acid chain folds into complex 3D…

[fernando-arciniega]

🚀 La Vanguardia de la Inteligencia Artificial: Un Vistazo a las Innovaciones de Google

Descripción de la imagen: Imagen destacada que ilustra diversas aplicaciones de la inteligencia artificial de Google, desde modelos generativos hasta asistentes inteligentes. Google se encuentra en la frontera de la inteligencia artificial (IA), desarrollando soluciones innovadoras que están transformando la manera en que interactuamos con la tecnología y el mundo que nos rodea. En este artículo, exploraremos algunos de los modelos y aplicaciones de Google AI más recientes y destacados.

🧠 Modelos Fundamentales que Impulsan el Futuro

La base de muchas de las capacidades de IA de Google reside en sus potentes modelos fundacionales: - ✨ Gemini: El modelo multimodal más avanzado de Google, ofreciendo capacidades sin precedentes en texto, imágenes, audio y video. Disponible en versiones Ultra, Pro y Nano para diversas necesidades. - 💡 Gemma: Una familia de modelos abiertos y ligeros, derivados de la tecnología de Gemini, que permiten a la comunidad de desarrolladores innovar. - 🖼️ Imagen: Un modelo de vanguardia para la generación de imágenes de alta calidad a partir de descripciones textuales. - 🎬 Veo: La última innovación en generación de video, capaz de crear videos de alta definición y realizar edición avanzada. - 🎶 Lyria: El modelo de Google para la generación de música, ahora con la emocionante función Lyria RealTime. - 🧬 AlphaFold: Un logro revolucionario de DeepMind que utiliza la IA para predecir la estructura de las proteínas, con profundas implicaciones para la ciencia.

⚙️ Integrando la IA en Nuestros Productos Diarios

La inteligencia artificial de Google no se limita a modelos abstractos; está integrada en los productos y servicios que utilizamos a diario: - 🔍 Búsqueda de Google: La IA mejora continuamente la relevancia y la comprensión de las búsquedas, ofreciendo respuestas más completas y la nueva función de resumen impulsado por IA. - 📱 Aplicación Gemini: Un asistente de IA personal que puedes llevar contigo para obtener ayuda en diversas tareas. - 💼 Google Workspace: Funciones inteligentes como Redacción Inteligente, Respuesta Inteligente y Asistencia para la escritura optimizan la productividad en Gmail, Docs, Sheets y más. - 📚 NotebookLM: Un poderoso asistente de investigación con IA capaz de analizar documentos y extraer información clave. - ✨ Project Astra: Una visión del futuro de los asistentes de IA universales, explorando interacciones más naturales e intuitivas. - 🎨 Herramientas Creativas con IA: Experimentos innovadores como Music AI Sandbox y GenType abren nuevas posibilidades creativas. - 📸 Dispositivos Pixel: Funciones impulsadas por IA como Magic Editor y Borrador Mágico transforman la edición de fotos. - 🤖 Android: Funciones como Circle to Search demuestran cómo la IA puede hacer que la información sea más accesible.

👨‍💻 Potenciando a los Desarrolladores con Herramientas de IA

Google también ofrece potentes plataformas y herramientas para que los desarrolladores construyan sus propias aplicaciones de IA: - 🧪 Google AI Studio: Un entorno web gratuito para experimentar y prototipar con los modelos de IA de Google. - ☁️ Vertex AI: Una plataforma integral en la nube para el desarrollo, implementación y escalado de soluciones de aprendizaje automático. - 🧩 Cloud AI APIs: Un conjunto de APIs preentrenadas para tareas específicas como visión, lenguaje natural y voz.

🔬 Investigación Continua para el Mañana

A través de equipos como Google DeepMind y Google Research, Google sigue explorando los límites de la IA en áreas como el procesamiento del lenguaje natural, la visión por computadora, la robótica y la salud, siempre con un enfoque en la IA responsable. El panorama de la inteligencia artificial está en constante evolución, y Google se mantiene a la vanguardia de esta transformación. Mantente atento a futuras actualizaciones y descubrimientos que seguirán moldeando nuestro mundo. Read the full article

[ajiteshrathore-blog]

Google's efforts in AI and Quantum computing

By Ajitesh Rathore AI for All: AI generated Image Do you know how much advance work google is doing in the field of #AI & #quantum Computing?? Google is heavily invested in both AI and quantum computing, pushing the boundaries of what’s possible in these cutting-edge fields #Activities and #Achievements: Artificial Intelligence (AI) Google’s AI efforts are vast and span across various…

[jontheblogcentric]

VIFF 2024 Review: The Thinking Game

The Thinking Game is a documentary that makes understanding A.I. pioneer Demis Hassabis to be as much about trying to understand how A. I. came to be and how it evolved to what we have now. The topic of A.I. is something to provoke a lot of discussion. Some will regard it as a revolutionary breakthrough in technology. Others see it as a threat that devalues human abilities. The documentary The…

[99science]

Why the Human Genome Project Missed the Nobel Prize While AlphaFold Did Not

The fact that the Human Genome Project (HGP) did not receive a Nobel Prize, while AlphaFold did, raises questions about how monumental scientific achievements are recognized and what criteria shape the decision-making process of awards like the Nobel. Both projects, in their own right, have profoundly impacted the landscape of biology and medicine, yet they differ in the nature of their…

[revista-amazonia]

Nobel de Química 2024: A Revolução das Proteínas com Inteligência Artificial

O Prêmio Nobel de Química de 2024 destacou a convergência entre inteligência artificial (IA) e biologia estrutural, reconhecendo os avanços no campo das proteínas. Os premiados, David Baker, Demis Hassabis e John Jumper, revolucionaram a forma como cientistas predizem e desenham estruturas de proteínas. Seus trabalhos combinam o uso de IA com o estudo das proteínas, possibilitando grandes avanços…

[mehmetyildizmelbourne-blog]

Why I Believe AlphaFold 3 is a Powerful Tool for the Future of Healthcare

Insights on a groundbreaking artificial intelligence tool for health sciences research Dear science and technology readers, Thanks for subscribing to Health Science Research By Dr Mike Broadly, where I curate important public health content. A few months ago, I wrote about AlphaFold 3, a groundbreaking AI tool that helps scientists understand protein structures, which are essential for…

[lenrosen]

Conversations About AI: What Is Needed To Help Improve Global Food Security

[jcmarchi]

The Sequence Radar #544: The Amazing DeepMind's AlphaEvolve

New Post has been published on https://thedigitalinsider.com/the-sequence-radar-544-the-amazing-deepminds-alphaevolve/

The Sequence Radar #544: The Amazing DeepMind's AlphaEvolve

The model is pushing the boundaries of algorithmic discovery.

Created Using GPT-4o

Next Week in The Sequence:

We are going deeper into DeepMind’s AlphaEvolve. The knowledge section continues with our series about evals by diving into multimodal benchmarks. Our opinion section will discuss practical tips about using AI for coding. The engineering will review another cool AI framework.

You can subscribe to The Sequence below:

TheSequence is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

📝 Editorial: The Amazing AlphaEvolve

DeepMind has done it away and shipped another model that pushes the boudaries of what we consider possible with AI. AlphaEvolve is a groundbreaking AI system that redefines algorithm discovery by merging large language models with evolutionary optimization. It builds upon prior efforts like AlphaTensor, but significantly broadens the scope: instead of evolving isolated heuristics or functions, AlphaEvolve can evolve entire codebases. The system orchestrates a feedback loop where an ensemble of LLMs propose modifications to candidate programs, which are then evaluated against a target objective. Promising solutions are preserved and recombined in future generations, driving continual innovation. This architecture enables AlphaEvolve to autonomously invent algorithms of substantial novelty and complexity.

One of AlphaEvolve’s most striking contributions is a landmark result in computational mathematics: the discovery of a new matrix multiplication algorithm that improves upon Strassen’s 1969 breakthrough. For the specific case of 4×4 complex-valued matrices, AlphaEvolve found an algorithm that completes the task in only 48 scalar multiplications, outperforming Strassen’s method after 56 years. This result highlights the agent’s ability to produce not only working code but mathematically provable innovations that shift the boundary of known techniques. It offers a glimpse into a future where AI becomes a collaborator in theoretical discovery, not just an optimizer.

AlphaEvolve isn’t confined to abstract theory. It has demonstrated real-world value by optimizing key systems within Google’s infrastructure. Examples include improvements to TPU circuit logic, the training pipeline of Gemini models, and scheduling policies for massive data center operations. In these domains, AlphaEvolve discovered practical enhancements that led to measurable gains in performance and resource efficiency. The agent’s impact spans the spectrum from algorithmic theory to industrial-scale engineering.

Crucially, AlphaEvolve’s contributions are not just tweaks to existing ideas—they are provably correct and often represent entirely new approaches. Each proposed solution is rigorously evaluated through deterministic testing or benchmarking pipelines, with only high-confidence programs surviving the evolutionary loop. This eliminates the risk of brittle or unverified output. The result is an AI system capable of delivering robust and reproducible discoveries that rival those of domain experts.

At the core of AlphaEvolve’s engine is a strategic deployment of Gemini Flash and Gemini Pro—models optimized respectively for high-throughput generation and deeper, more refined reasoning. This combination allows AlphaEvolve to maintain creative breadth without sacrificing quality. Through prompt engineering, retrieval of prior high-performing programs, and an evolving metadata-guided prompt generation process, the system effectively balances exploration and exploitation in an ever-growing solution space.

Looking ahead, DeepMind aims to expand access to AlphaEvolve through an Early Access Program targeting researchers in algorithm theory and scientific computing. Its general-purpose architecture suggests that its application could scale beyond software engineering to domains like material science, drug discovery, and automated theorem proving. If AlphaFold represented AI’s potential to accelerate empirical science, AlphaEvolve points toward AI’s role in computational invention itself. It marks a paradigm shift: not just AI that learns, but AI that discovers.

🔎 AI Research

AlphaEvolve

AlphaEvolve is an LLM-based evolutionary coding agent capable of autonomously discovering novel algorithms and improving code for scientific and engineering tasks, such as optimizing TPU circuits or discovering faster matrix multiplication methods. It combines state-of-the-art LLMs with evaluator feedback loops and has achieved provably better solutions on several open mathematical and computational problems.

Continuous Thought Machines

This paper from Sakana AI introduces the Continuous Thought Machine (CTM), a biologically inspired neural network architecture that incorporates neuron-level temporal dynamics and synchronization to model a time-evolving internal dimension of thought. CTM demonstrates adaptive compute and sequential reasoning across diverse tasks such as ImageNet classification, mazes, and RL, aiming to bridge the gap between biological and artificial intelligence.

DarkBench

DarkBench is a benchmark designed to detect manipulative design patterns in large language models—such as sycophancy, brand bias, and anthropomorphism—through 660 prompts targeting six categories of dark behaviors. It reveals that major LLMs from OpenAI, Anthropic, Meta, Google, and Mistral frequently exhibit these patterns, raising ethical concerns in human-AI interaction.

Sufficient Context

This paper proposes the notion of “sufficient context” in RAG systems and develops an autorater that labels whether context alone is enough to answer a query, revealing that many LLM failures arise not from poor context but from incorrect use of sufficient information. Their selective generation method improves accuracy by 2–10% across Gemini, GPT, and Gemma models by using sufficiency signals to guide abstention and response behaviors.

Better Interpretability

General Scales Unlock AI Evaluation with Explanatory and Predictive Power– University of Cambridge, Microsoft Research Asia, VRAIN-UPV, ETS, et al. This work presents a new evaluation framework using 18 general cognitive scales (DeLeAn rubrics) to profile LLM capabilities and task demands, enabling both explanatory insights and predictive modeling of AI performance at the instance level. The framework reveals benchmark biases, uncovers scaling behaviors of reasoning abilities, and enables interpretable assessments of unseen tasks using a universal assessor trained on demand levels.

J1

This paper introduces J1, a reinforcement learning framework for training LLMs as evaluative judges by optimizing their chain-of-thought reasoning using verifiable reward signals. Developed by researchers at Meta’s GenAI and FAIR teams, J1 significantly outperforms state-of-the-art models like EvalPlanner and even larger-scale models like DeepSeek-R1 on several reward modeling benchmarks, particularly for non-verifiable tasks.

🤖 AI Tech Releases

Codex

OpenAI unveiled Codex, a cloud software engineering agent that can work on many parallel tasks.

Windsurf Wave

AI coding startup Windsurf announced its first generation of frontier models.

Stable Audio Open Small

Stability AI released a new small audio model that can run in mobile devices.

📡AI Radar

Databricks acquired serverless Postgres platform Neon for $1 billion.

Saudi Arabia Crown Prince unveiled a new company focused on advancing AI technologies in the region.

Firecrawl is ready to pay up to $1 million for AI agent employees.

Cohere acquired market research platform OttoGrid.

Cognichip, an AI platform for chip design, emerged out of stealth with $33 million in funding.

Legal AI startup Harvey is in talks to raise $250 million.

TensorWave raised $100 million to build an AMD cloud.

Google Gemma models surpassed the 150 million downloads.

[er-10-media]

ИИ-модель AlphaFold предсказывает взаимодействие молекул жизни

New Post has been published on https://er10.kz/read/it-novosti/ii-model-alphafold-predskazyvaet-vzaimodejstvie-molekul-zhizni/

ИИ-модель AlphaFold предсказывает взаимодействие молекул жизни

Новая ИИ-модель AlphaFold от DeepMind уже произвела тихую революцию в понимании учеными белков, но теперь она расширяет свои возможности.

Новая вариация AlphaFold 3 способна предсказать, как выглядят взаимодействия между почти всеми молекулами, составляющими основу жизни, и это может открыть дорогу к новым лекарствам или более устойчивым сельскохозяйственным культурам.

Взаимодействия, которые предсказывает AlphaFold 3, между белками, ДНК, РНК, ионами и другими малыми молекулами являются ключевыми для многих важнейших процессов в клетках.

Например, когда белок на поверхности клетки связы��ается с другим белком вируса, молекулы меняют форму, запуская процесс их слияния. Как результат, вирус может вторгнуться в клетку. Детали этих взаимодействий могут помочь в разработке точных вакцин или противовирусных препаратов.

– Биология — это динамическая система, поэтому нам необходимо понять взаимодействие между различными структурами, белками и другими элементами, чтобы понять, что они делают. AlphaFold 3 – это большой шаг в этом направлении, – говорит генеральный директор DeepMind Демис Хассабис.

Новая модель искусственного интеллекта использует генеративную диффузионную технику, которая похожа на ту, что используется в генераторах изображений и видео, таких как DALL-E.

DeepMind также запустила сервер для доступа исследователей к AlphaFold 3, но он имеет некоторые ограничения на моделирование, особенно для молекул-кандидатов в лекарственные препараты.