MAROKO133 Hot ai: US firm’s drones with 1,000 lbs of suppression power bring faster fire r

📌 MAROKO133 Hot ai: US firm’s drones with 1,000 lbs of suppression power bring fas

A US startup has secured funding to scale the production of its autonomous firefighting drone technology to tackle rising wildfire incidents.

Seattle-based Seneca is developing the first fully autonomous fire suppression system featuring drones that can deliver fire-suppressing agents at high pressure and use AI to navigate and extinguish fires in under ten minutes.

These modular aerial units can be launched remotely with an approximate fire location, striking quickly during the critical window when a spark can turn into a megafire.

The startup has now announced its public launch and a $60 million funding round, led by Caffeinated Capital and Convective Capital—the largest in fire tech to date.

“Our hope is to empower firefighters in situations that were previously impossible, unsafe, or inefficient. The American way of life has always been about pioneering. Technology is how we will protect communities, preserve our environment, and create a more resilient civilization,” said Stuart Landesberg, Founder and CEO of Seneca, in a statement.

Drones battle wildfires

Wildfire intensity in the United States has more than doubled over the past two decades, inflicting an estimated $1 trillion in annual economic losses and endangering 115 million Americans.

Experts warn that 2025 will become the most destructive and costly fire year in US history. The devastation extends far beyond the loss of homes, impacting the air people breathe, the water they depend on, and the forests vital to the nation’s environment and way of life.

Amid the highest wildfire risk in history, Seneca aims to modernize firefighting infrastructure with advanced autonomous drone technology, AI, and computer vision. Developed with insights alongside firefighters and landowners, Seneca claims its solutions are built to integrate seamlessly into the operations of leading fire agencies, utilities, and property owners.

The firm’s modular aerial suppression units are compact enough to be hand-carried, transported in utility vehicles, or deployed remotely. This flexibility enhances efficiency and effectiveness across multiple scenarios, from verifying and containing early-stage fires in remote terrain to supporting large-scale firefighting and cold trailing efforts on major incidents.

“Seneca’s vision for rapid, drone-based response is a critical missing capability that allows firefighters to conduct suppression operations when it is still feasible to do so. This should be considered essential for stakeholders ranging from fire agencies to utilities, to municipalities; the opportunity for impact is enormous,” said Bill Clerico, Managing Partner of Convective Capital, in a statement.

AI fire resilience

Seneca’s autonomous firefighting system represents a breakthrough in aerial fire suppression technology, combining robotics, artificial intelligence, and advanced computer vision to respond faster and more safely than traditional methods.

The modular system centers on high-performance electric drones capable of carrying over 100 pounds (45 kilograms) of fire-suppressing agents and delivering them at ultra-high PSI using aerated Class A foam for rapid knockdown.

Designed to operate independently or in coordinated swarms, groups of four to six drones can deliver 500 to 1,000 pounds (227 to 454 kilograms) of suppression power per mission—without needing helipads, refueling stations, or complex ground infrastructure.

Operating from a portable tablet, the system can reach hazardous or isolated areas that human personnel cannot, and it can launch in a matter of seconds, even if it is only given an estimated fire location. Seneca drones, which compensate for wind and visibility, can precisely detect, track, and target fires because of their AI-driven navigation and onboard infrared sensors.

The company claims that camera-based situational awareness, ADS-B, and Remote ID integration, and autonomous obstacle avoidance contribute to safety and guarantee smooth operation in challenging skies. The platform offers unparalleled flexibility for year-round fire management by supporting planned burns, structural protection, hazardous control, utility safety activities, and emergency response.

Developed in collaboration with firefighters and landowners, Seneca claims its scalable aerial units will redefine how wildfires are detected, controlled, and contained in an era of growing climate risk.

🔗 Sumber: interestingengineering.com

📌 MAROKO133 Update ai: MIT Researchers Unveil “SEAL”: A New Step Towards Self-Impr

The concept of AI self-improvement has been a hot topic in recent research circles, with a flurry of papers emerging and prominent figures like OpenAI CEO Sam Altman weighing in on the future of self-evolving intelligent systems. Now, a new paper from MIT, titled “Self-Adapting Language Models,” introduces SEAL (Self-Adapting LLMs), a novel framework that allows large language models (LLMs) to update their own weights. This development is seen as another significant step towards the realization of truly self-evolving AI.

The research paper, published yesterday, has already ignited considerable discussion, including on Hacker News. SEAL proposes a method where an LLM can generate its own training data through “self-editing” and subsequently update its weights based on new inputs. Crucially, this self-editing process is learned via reinforcement learning, with the reward mechanism tied to the updated model’s downstream performance.

The timing of this paper is particularly notable given the recent surge in interest surrounding AI self-evolution. Earlier this month, several other research efforts garnered attention, including Sakana AI and the University of British Columbia’s “Darwin-Gödel Machine (DGM),” CMU’s “Self-Rewarding Training (SRT),” Shanghai Jiao Tong University’s “MM-UPT” framework for continuous self-improvement in multimodal large models, and the “UI-Genie” self-improvement framework from The Chinese University of Hong Kong in collaboration with vivo.

Adding to the buzz, OpenAI CEO Sam Altman recently shared his vision of a future with self-improving AI and robots in his blog post, “The Gentle Singularity.” He posited that while the initial millions of humanoid robots would need traditional manufacturing, they would then be able to “operate the entire supply chain to build more robots, which can in turn build more chip fabrication facilities, data centers, and so on.” This was quickly followed by a tweet from @VraserX, claiming an OpenAI insider revealed the company was already running recursively self-improving AI internally, a claim that sparked widespread debate about its veracity.

Regardless of the specifics of internal OpenAI developments, the MIT paper on SEAL provides concrete evidence of AI’s progression towards self-evolution.

Understanding SEAL: Self-Adapting Language Models

The core idea behind SEAL is to enable language models to improve themselves when encountering new data by generating their own synthetic data and optimizing their parameters through self-editing. The model’s training objective is to directly generate these self-edits (SEs) using data provided within the model’s context.

The generation of these self-edits is learned through reinforcement learning. The model is rewarded when the generated self-edits, once applied, lead to improved performance on the target task. Therefore, SEAL can be conceptualized as an algorithm with two nested loops: an outer reinforcement learning (RL) loop that optimizes the generation of self-edits, and an inner update loop that uses the generated self-edits to update the model via gradient descent.

This method can be viewed as an instance of meta-learning, where the focus is on how to generate effective self-edits in a meta-learning fashion.

A General Framework

SEAL operates on a single task instance (C,τ), where C is context information relevant to the task, and τ defines the downstream evaluation for assessing the model’s adaptation. For example, in a knowledge integration task, C might be a passage to be integrated into the model’s internal knowledge, and τ a set of questions about that passage.

Given C, the model generates a self-edit SE, which then updates its parameters through supervised fine-tuning: θ′←SFT(θ,SE). Reinforcement learning is used to optimize this self-edit generation: the model performs an action (generates SE), receives a reward r based on LMθ′’s performance on τ, and updates its policy to maximize the expected reward.

The researchers found that traditional online policy methods like GRPO and PPO led to unstable training. They ultimately opted for ReST^EM, a simpler, filtering-based behavioral cloning approach from a DeepMind paper. This method can be viewed as an Expectation-Maximization (EM) process, where the E-step samples candidate outputs from the current model policy, and the M-step reinforces only those samples that yield a positive reward through supervised fine-tuning.

The paper also notes that while the current implementation uses a single model to generate and learn from self-edits, these roles could be separated in a “teacher-student” setup.

Instantiating SEAL in Specific Domains

The MIT team instantiated SEAL in two specific domains: knowledge integration and few-shot learning.

• Knowledge Integration: The goal here is to effectively integrate information from articles into the model’s weights.
• Few-Shot Learning: This involves the model adapting to new tasks with very few examples.

Experimental Results

The experimental results for both few-shot learning and knowledge integration demonstrate the effectiveness of the SEAL framework.

In few-shot learning, using a Llama-3.2-1B-Instruct model, SEAL significantly improved adaptation success rates, achieving 72.5% compared to 20% for models using basic self-edits without RL training, and 0% without adaptation. While still below “Oracle TTT” (an idealized baseline), this indicates substantial progress.

For knowledge integration, using a larger Qwen2.5-7B model to integrate new facts from SQuAD articles, SEAL consistently outperformed baseline methods. Training with synthetically generated data from the base Qwen-2.5-7B model already showed notable improvements, and subsequent reinforcement learning further boosted performance. The accuracy also showed rapid improvement over external RL iterations, often surpassing setups using GPT-4.1 generated data within just two iterations.

Qualitative examples from the paper illustrate how reinforcement learning leads to the generation of more detailed self-edits, resulting in improved performance.

While promising, the researchers also acknowledge some limitations of the SEAL framework, including aspects related to catastrophic forgetting, computational overhead, and context-dependent evaluation. These are discussed in detail in the original paper.

Original Paper: https://arxiv.org/pdf/2506.10943

Project Site: https://jyopari.github.io/posts/seal

Github Repo: https://github.com/Continual-Intelligence/SEAL

The post MIT Researchers Unveil “SEAL”: A New Step Towards Self-Improving AI first appeared on Synced.

🔗 Sumber: syncedreview.com