video prediction
30 articles about video prediction in AI news
Dyna-2 World-Action Model Trained on 1M Hours Video
Dyna-2, trained on 1M+ hours of egocentric video, jointly predicts future video and actions. Claims new scaling laws but no benchmarks or technical details released.
GenCeption: Video Diffusion Backbone Beats Specialists on 5 Vision Tasks
GenCeption uses video diffusion as a vision backbone, matching specialists with 7-500x less data and generalizing from synthetic to real footage.
Elon Musk Predicts 'Vast Majority' of AI Compute Will Be for Real-Time Video
Elon Musk states that real-time video consumption and generation will consume most AI compute, highlighting a shift from text to video as the primary medium for AI processing.
Meta's V-JEPA 2.1 Achieves +20% Robotic Grasp Success with Dense Feature Learning from 1M+ Hours of Video
Meta researchers released V-JEPA 2.1, a video self-supervised learning model that learns dense spatial-temporal features from over 1 million hours of video. The approach improves robotic grasp success by ~20% over previous methods by forcing the model to understand precise object positions and movements.
Beyond Simple Recognition: How DeepIntuit Teaches AI to 'Reason' About Videos
Researchers have developed DeepIntuit, a new AI framework that moves video classification from simple pattern imitation to intuitive reasoning. The system uses vision-language models and reinforcement learning to handle complex, real-world video variations where traditional models fail.
BAAI Orca World Model Matches π0.5 With No Action Labels
BAAI's Orca world model matches specialized π0.5 on five robotics tasks, trained on 125,000 hours of video without action labels, predicting abstract world states.
World Action Models Survey Unifies 100+ Methods Under One Taxonomy
A survey reviews 100+ world action models, unifying world models, video generation, and VLA policies under one taxonomy.
mlx-vlm v0.5.0 Adds Continuous Batching, Distributed Inference for Apple Silicon
mlx-vlm v0.5.0 adds continuous batching, speculative decoding, and distributed inference for Apple Silicon. The release supports Qwen3.5, Kimi K2.5, Gemma 4 video, and new models with 21 contributors.
New Research Proposes CPGRec
A new arXiv paper introduces CPGRec, a three-module framework for video game recommendations. It aims to solve the common trade-off between accuracy and diversity by using strict game connections and leveraging category/popularity data. Experiments on a Steam dataset show promising results.
Developer Swaps Dash Cam Analysis for Gemma 4 & Falcon Perception
A developer announced they are replacing their entire dash cam video analysis system with Google's Gemma 4 and Falcon Perception models, signaling a practical shift towards newer, specialized multimodal models for real-time edge applications.
PilotBench Exposes LLM Physics Gap: 11-14 MAE vs. 7.01 for Forecasters
PilotBench, a new benchmark built from 708 real-world flight trajectories, evaluates LLMs on safety-critical physics prediction. It uncovers a 'Precision-Controllability Dichotomy': LLMs follow instructions well but suffer high error (11-14 MAE), while traditional forecasters are precise (7.01 MAE) but lack semantic reasoning.
Sundar Pichai Predicts 'Profound' AI Shifts for 2027
Google CEO Sundar Pichai stated he expects 2027 to be a 'big year' where major AI shifts happen 'pretty profoundly.' This is a specific, forward-looking prediction from a key industry leader about the pace of change.
AI Forecasters Revise AGI Timeline: Key Milestones Pulled Forward to 2029-2030 After Recent Model Progress
A significant update from AI forecasters indicates key AGI milestones have been pulled forward, with the median prediction for AGI arrival shifting from 2032 to 2029-2030. This revision follows rapid progress in recent model capabilities, particularly in reasoning and tool use.
Nvidia Claims MLPerf Inference v6.0 Records with 288-GPU Blackwell Ultra Systems, Highlights 2.7x Software Gains
MLCommons released MLPerf Inference v6.0 results, introducing multimodal and video model tests. Nvidia set records using 288-GPU Blackwell Ultra systems and achieved a 2.7x performance jump on DeepSeek-R1 via software optimizations alone.
Zero-Shot Cross-Domain Knowledge Distillation: A YouTube-to-Music Case Study
Google researchers detail a case study transferring knowledge from YouTube's massive video recommender to a smaller music app, using zero-shot cross-domain distillation to boost ranking models without training a dedicated teacher. This offers a practical blueprint for improving low-traffic AI systems.
LeCun's Team Publishes LeWorldModel: A 15M-Parameter World Model That Mathematically Prevents Training Collapse
Yann LeCun's team has open-sourced LeWorldModel, a 15M-parameter world model that uses a novel SIGReg regularizer to make representation collapse mathematically impossible. It trains on a single GPU in hours and enables efficient physical prediction for robotics and autonomous systems.
The Digital Twin Revolution: How LLMs Are Creating Virtual Testbeds for Social Media Policy
Researchers have developed an LLM-augmented digital twin system that simulates short-video platforms like TikTok to test policy changes before implementation. This four-twin architecture allows platforms to study long-term effects of AI tools and content policies in realistic closed-loop simulations.
NVIDIA's DreamDojo: Teaching Robots to 'Dream' in Pixels with 44,000 Hours of Human Experience
NVIDIA has open-sourced DreamDojo, a revolutionary robot world model trained on 44,711 hours of real-world human video. Instead of relying on physics engines, it predicts action outcomes directly in pixel space, potentially accelerating robotics development by orders of magnitude.
Scaling Laws Differ for Native Multimodal VLMs
A systematic study reveals distinct scaling laws for native multimodal pre-training, showing vision-language models require different compute-optimal size/token ratios than language-only models.
Xiaomi Open-Sources 38B Robotics-U0 Unifying Four Embodied Tasks
Xiaomi open-sourced 38B-parameter Robotics-U0, unifying four embodied tasks in a single model. No benchmarks or training data disclosed yet.
Hugging Face Papers: 35B Agent Matches Trillion-Parameter Performance
Hugging Face Daily Papers featured eight AI papers, including Orca (world model), Dockerless (62% SWE-bench), and a 35B agent matching trillion-parameter performance.
Karpathy: Neural nets will become the host, CPUs the co-processor
Karpathy predicts neural networks will become the host OS, with CPUs as co-processors, rendering most classical app interfaces obsolete.
Meta Deploys Millions of Amazon Graviton CPUs for AI Agents
Meta will deploy tens of millions of AWS Graviton5 CPU cores for AI agent workloads, signaling that agentic inference favors CPUs over GPUs. The deal deepens Meta's $200B+ infrastructure push amid layoffs and cloud rivalry.
CGCMA Model Achieves +0.449 Sharpe Ratio in Asynchronous Crypto News Fusion
Researchers propose CGCMA, a model for fusing sporadic news with continuous market data. It achieved a +0.449 Sharpe ratio on a new crypto trading benchmark, showing gains not explained by simple heuristics.
Yann LeCun's JEPA Vision Gains Traction as Generative AI Hits Limits
A widely-shared critique claims the generative AI paradigm is a dead end, aligning with Meta's Yann LeCun's years of advocating for his Joint Embedding Predictive Architecture (JEPA) approach.
AI-Powered PS4 Emulator 'Spine' Runs Bloodborne Locally on PC
A developer has released Spine, a PS4 emulator that uses AI techniques to run Bloodborne fully on PC. This represents a major step forward in console emulation, previously considered years away.
Mark Cuban Predicts AI Integration Wave for 33M US SMBs
Mark Cuban predicts the next major job wave will be in custom AI integration for small to mid-sized companies, stating generic 'software is dead' as everything becomes uniquely customized. He highlights a market of 33 million US companies needing these services.
Tencent Open-Sources HY-World 2.0 Multimodal 3D World Model
Tencent's Hunyuan AI lab has open-sourced HY-World 2.0, a multimodal world model capable of generating, reconstructing, and simulating interactive 3D scenes. This release provides a significant, freely available tool for 3D content creation and embodied AI research.
Google Open-Sources Magika AI for File Detection, 99% Accuracy at 5ms
Google released Magika, an AI model trained on 100M files to identify over 200 content types with 99% accuracy in 5ms. It was Google's internal 'secret weapon' for years, now available via pip install.
AllenAI's WildDet3D Enables Promptable 3D Object Detection from Single Images
Allen Institute for AI (AllenAI) has open-sourced WildDet3D, a model for promptable 3D object detection from single RGB images. It predicts 3D bounding boxes using flexible prompts and can integrate optional depth data.