Search the archive

market data forecasting model. Registration is open.

GOOGLE'S HEAD OF RESEARCH ON WHETHER 'LEARN TO CODE' IS STILL GOOD ADVICE IN THE AGE OF AI

Despite AI handling some coding tasks, understanding coding basics is crucial and creates new opportunities across fields, including healthcare and education.

WHY ARE WE USING LLMS AS CALCULATORS?

Researchers are experimenting with LLMs' ability to solve math problems to assess their reasoning capabilities.

GPTS ARE MAXED OUT

OpenAI's next-gen model, internally known as Orion, reportedly underperforms compared to expectations set by Sam Altman, suggesting a potential limit to scalable growth in AI models.

INSIDE OPENAI'S EARLY DRAMA

A lawsuit by Elon Musk against OpenAI has resulted in emails from the startup's early days being revealed, shedding light on internal tensions.

ANTHROPIC HIRES ITS FIRST “AI WELFARE” RESEARCHER

Anthropic hired Kyle Fish as its first "AI welfare" researcher to explore ethical considerations around potential AI consciousness and moral rights. This marks a potential shift for AI companies in addressing ethical questions related to AI systems' consciousness and agency. Fish's recent paper discusses improving AI welfare understanding to avoid mishandling decisions concerning AI moral consideration.

SPOTIFY'S PLANS FOR AI GENERATED MUSIC, PODCASTS, AND RECOMMENDATIONS, ACCORDING TO ITS CO-PRESIDENT, CTO, AND CPO GUSTAV SÖDERSTRÖM

Spotify is embracing AI-powered content creation, using generative AI tools like Suno and NotebookLM for music and podcasts while leveraging LLMs to enhance user recommendations. Co-President Gustav Söderström says AI is a tool for amplifying creativity rather than replacing it and that AI has the potential to foster deeper user engagement and personalized experiences. Spotify remains committed to supporting creators on its platform, ensuring legal compliance while exploring dynamic AI-driven innovation.

LLAMA MESH

Fine-tuned model from Nvidia based on LlaMA that can generate vertices for 3D models based on textual input.

SEMANTIC AWARE WATERMARKING

Researchers introduce the Semantic Perturbation Attack to counter existing attacks on watermarking schemes.

SOLVING THE TRAVELLING SALESMAN PROBLEM

This study reveals the overlooked importance of the “heatmap + Monte Carlo Tree Search (MCTS)” approach, showing that careful tuning and simple heatmaps can outperform complex models.

RUN LENGTH TOKENIZATION FOR VIDEO TRANSFORMERS

You can dramatically speed up run time with no loss in performance without additional training by adaptively caching video tokens that have not changed over frames.

REGION AWARE TEXT-TO-IMAGE GENERATION

An improved technique for generating images with improved control based on chosen regions.

ACCURATE IMAGE MATCHING

MOP+MiHo+NCC is a non-deep, modular approach for refining image matches by combining three techniques. Multiple Overlapping Planes (MOP) clusters inlier matches and filters outliers with RANSAC. Middle Homography (MiHo) reduces distortion in planar reprojection. Normalized Cross Correlation (NCC) refines keypoint positions after transformation.

GRAPH-BASED AI MODEL MAPS THE FUTURE OF INNOVATION

MIT researchers developed an AI model using generative knowledge extraction and graph reasoning to uncover complex patterns linking diverse fields like biology and music. The model efficiently creates knowledge maps from scientific papers, identifying connections and suggesting innovative materials inspired by art. This approach enhances interdisciplinary research, revealing hidden insights and novel concepts for material design.

THE BEGINNER'S GUIDE TO VISUAL PROMPT INJECTIONS

WHAT IF AI DOESN'T JUST KEEP GETTING BETTER FOREVER?

Reports indicate that traditional LLM training may be hitting diminishing returns, as new models like OpenAI's Orion are not significantly outperforming predecessors. Experts are concerned about running out of quality textual data for LLM training, prompting shifts towards synthetic data and specialized AI models. Future advancements might focus on reasoning improvements and task-specific models rather than general scaling.

and join →

AUTOMATED PYTHON PACKAGE CREATION

Pygen turns ideas into Python Packages.

ULTRAVOX AUDIO LANGUAGE MODELS (HUGGING FACE HUB)

A suite of open weight models that can take text and audio as input modalities.

AI MAKES TECH DEBT MORE EXPENSIVE

AI amplifies the cost of tech debt by widening the velocity gap between low-debt and high-debt codebases.

experiences like virtual try-ons. This eBook from Algolia

clarifies the differences and synergies between genAI and LLMs and reveals how these technologies can streamline operations while maximizing customer engagement and satisfaction. Download your free copy today [4] to uncover how AI can set your ecommerce business apart in today's competitive landscape.

FRANCOIS CHOLLET LEAVES GOOGLE

The creator of Keras and the Arc eval (among many other things) has left Google. He will continue to support the Jax and Keras community while pursuing other endeavors.

CHATGPT DESKTOP

OpenAI has launched ChatGPT desktop integrations to rival Copilot.

OPENAI TO PRESENT PLANS FOR U.S. AI STRATEGY AND AN ALLIANCE TO COMPETE WITH CHINA

OpenAI's AI infrastructure blueprint proposes AI economic zones and partnerships with the U.S. Navy on nuclear power to boost AI-driven economic growth and innovation. The plan includes a North American AI alliance and initiatives similar to the National Interstate and Defense Highways Act to support infrastructure needs. It emphasizes investment in U.S. data centers and energy projects to maintain a competitive edge over China.

INTRODUCING ATHENE-V2: ADVANCING BEYOND THE LIMITS OF SCALING WITH TARGETED POST-TRAINING

Athene V2 is a series of models trained on top of Qwen 2.5 72B. They are specifically tuned for agentic and chat workflows and surpass GPT-4o in a number of critical benchmarks.

CONVOLUTIONAL DIFFERENTIABLE LOGIC GATE NETWORKS

PHYSICS-INFORMED DISTILLATION

Physics Informed Distillation (PID) is a technique that uses a student model to simplify and speed up diffusion models by representing them as solutions to differential equations.

AI value today. Read the blog by Camunda CEO Jakob Freund

MINERU: HIGH-QUALITY DATA EXTRACTION TOOL

MinerU is a high quality tool based on StructTable-InternVL2-1B that allows for the extraction of information from PDFs to a number of machine readable formats.

ISOTONIC REGRESSION

A powerful technique for fitting a monotonic function to data. It can be differentiated really well for a number of applications outside of curve fitting.

TEXT-TO-SQL QUERY

XiYan-SQL is a novel framework designed to improve the accuracy and diversity of SQL queries generated from natural language.

X PORTRAIT 2

Bytedance's AI group continues to release some amazing technology. X Portrait 2 is a continuation of its video acting model with improved realism and consistency for actor-to-video.

CREATE 360° VIEWS FROM LIMITED IMAGES

The MVSplat360 model offers a new way to create realistic 360° views of real-world scenes, even from just a few sparse images.

BRAIN TUMOR DETECTION

This paper showcases the top solution for brain tumor segmentation in the BraTS challenge, highlighting how synthetic data can enhance AI models for medical imaging tasks.

FLORA CREATIVE PLATFORM (PRODUCT)

AI Creative Platform FLORA has launched all 4 of the best Image to Video models (Runway, Luma, Kling, and Hailuo) in one tool.

SUPERMAVEN JOINS CURSOR

The code editing plugin's team is joining Cursor to continue developing the user experience.

GOOGLE'S AI ‘LEARNING COMPANION' TAKES CHATBOT ANSWERS A STEP FURTHER

Google's Learn About site provides answers with visual and interactive elements and educational formatting.

OPENAI'S AI AGENT MAY LAUNCH IN JANUARY

OpenAI is reportedly gearing up to release "Operator," an AI agent tool, as soon as January. According to Bloomberg, Operator could directly perform tasks on a user's computer. It will become available first as a research preview via OpenAI's developer API.

GOOGLE'S AI INITIATIVE TO BOOST MENA ECONOMY BY $320 BILLION

The Middle East and North Africa (MENA) could see a $320 billion economic boost from AI by 2030. To support this growth, Google.org has launched the AI Opportunity Initiative, its largest AI investment in the region. The initiative is focused on developing essential AI skills, funding research, and expanding AI access.

AI PROTEIN-PREDICTION TOOL ALPHAFOLD3 IS NOW OPEN SOURCE

DeepMind has released the code for AlphaFold3, allowing academic scientists to model protein-drug interactions and enhancing drug discovery efforts. Competitors like Baidu and ByteDance have developed similar tools, though commercial usage remains restricted. Ongoing development of fully open-source versions like OpenFold3 aims to bypass these limitations.

TWO TRILLION TOKEN COMMON CORPUS

This release marks the largest open and permissively licensed dataset for language model training. It has substantial multilingual representation.

THE RHYTHM IN ANYTHING

Zero shot drum style transfer for any input rhythm is a fun music use case for artists. This is done with a masked token modeling objective which is well suited for audio.

3D RECONSTRUCTION FOR DYNAMIC SCENES

HiCoM advances real-time 3D reconstruction from multi-view streaming videos by addressing challenges in storage, training speed, and rendering quality.