Tag
Box CEO Aaron Levie argues we're entering a new era of AI agents that take high-level goals, design solutions, and continuously optimize them, with a quoted timeline framing 2027 as the year of 'outcome engineering'.
CAL FIRE's Fire Innovation Unit, FireWERX, and Seneca demonstrated next-generation autonomous drone (UAS) technology in California that could transform wildfire response operations.
An AI agent autonomously planned and executed stereotaxic brain surgery on an epilepsy mouse model, selecting a gene therapy vector, brain region, and injection coordinates while a licensed vet supervised irreversible steps with a kill switch. The author teases a follow-up thread detailing the agent's decisions and a moment it refused an operator override.
Aaron Levie argues that autonomous, domain-specific agents represent the emerging form factor for AI, driven by better tool/computer use, improving model capabilities, and mature agent harnesses and infrastructure — positioning them as a major leap beyond simple LLM wrappers for enterprise workflows.
对比 Mythos 由40个层级化AI智能体组成的编队与GLM 400个智能体集群在第三方环境中的对抗推演,探讨多智能体协作与自主AI系统的能力差异。
A user utilized Opus 5.5 to automate clearing a 3D printer's print bed for non-stop experimentation without human intervention, as shown in a video.
A British startup, Worldmodeldata, is using video game data to train AI world models, aiming to overcome data shortages for teaching AI to understand and interact with the physical world.
Former Ukrainian defence minister Fedorov has launched an Army of Robots initiative aiming to deploy humanoids on the battlefield, with an AI-generated launch video. The initiative raises questions about the implications of autonomous machines in warfare.
Claude Opus 5.5 demonstrated autonomous control of a robot arm to replicate Michelangelo's work and independently noticed and corrected a mistake.
This paper proposes an LLM-driven unified conversion framework that enables automatic and semantically consistent transformation between behavior trees and finite state machines in autonomous intelligent systems, improving scalability and maintainability.
TechCrunch Disrupt 2026 features a panel discussion with leaders from Shield AI, Waabi, and General Motors on building AI systems for critical applications where failure has serious real-world consequences, emphasizing safety, validation, and trust.
The XPRIZE Wildfire competition awarded winners for rapid wildfire detection from space and autonomous response, but no team fully met the challenge of stopping fires within 10 minutes, highlighting advancements in detection technology while underscoring ongoing difficulties in fire suppression.
Earendil Robotics is developing autonomous interceptor drones and software to provide affordable and scalable air defense solutions, reducing the need for extensive pilot training.
Bernt Bornich, founder and CEO of 1X Tech, will speak at the Runtime event about the creation of the NEO Home Robot and his vision for a human-centered future with autonomous systems.
U.S. Chairman of the Joint Chiefs Gen. Dan Caine states that U.S. forces must prepare for being 'hunted' by autonomous systems in future warfare, citing AI-enabled drones in Ukraine and recent military operations as examples.
GPT-6 Astra autonomously completed a pen-spinning task using reinforcement learning with Isaac Lab and Sharpa hand, delivering a trained policy and visualization after a day and a half of runtime.
A NATO-backed startup, Scaleout Systems, is using small AI models and federated learning to enable drones to autonomously identify and attack targets in battlefield settings, leveraging edge computing for real-time updates.
This paper proposes a composite-gradient learning method that integrates deep reinforcement learning and model predictive control for shared control authority in autonomous systems, with evaluations on traffic networks showing modest benefits under strong interaction.
The paper introduces SELF-INDEX, a framework that enables search indexes to self-evolve autonomously, improving retrieval performance and benefiting downstream applications such as search agents and agent memory systems.
This paper presents a model-based reinforcement learning approach for autonomous droplet navigation in microfluidic platforms, achieving reliable navigation through complex geometries with zero-shot transfer from simpler paths.