MAROKO133 Update ai: MIT Researchers Unveil “SEAL”: A New Step Towards Self-Improving AI H

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

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

📌 MAROKO133 Eksklusif ai: SpaceX Hit With Back to Back Lawsuits From Workers Who S

SpaceX is facing a one-two punch of personal injury lawsuits after two different workers have sued the Elon Musk-owned company this month over being injured on the job, the San Antonio Express-News reports, adding to the poor safety track record at its Starbase facility in South Texas.

The latest suit, filed in Cameron County on Monday, was leveled against SpaceX and the steel firm W&W Erectors LLC by a subcontractor named Julian Escalante, who was working on one of the launchpads used by Starship — the largest rocket in the world, whose development is under massive pressure for being behind schedule.

According to the suit, Escalante’s right arm got “entangled and pinched” by a metal bucket holding “approximately 200 pounds” of industrial-sized bolts after it tumbled from a pallet.

“As the bucket fell, (Escalante’s) right arm was dragged downward with (the bucket),” the suit said, per the Express-News “The downward force pulled (his) right shoulder, and (he) fell with the bucket as it hit the ground.”

Just as concerning as the accident, which took place in November, was his higher ups’ alleged response to it. When Escalante reported it to his supervisor, the supervisor told him “not to report the injury and instructed him to return to work,” the suit claims, and his foreman, Joe Pedroza, had basically the same advice: “Just don’t tell anyone.”

But Escalante wanted medical care for his injury. His inquiries into where management stood on his request allegedly didn’t fly well with the General Foreman, identified only as “Wero,” who told Escalante to “be a man” and “stop crying,” according to the suit. His lawyers say that SpaceX was negligent in maintaining a safe job site.

A similar lawsuit was filed earlier this month by a worker named Sergio Ortiz, the Express-News reported, who says that in 2024 he was struck by falling debris while working in an elevator shaft at Starbase. According to the suit, the heavy cables used by welding machines called welding leads, which can weigh up to 80 pounds, fell from above and slammed his head.

The suits are the latest to put the safety track record of SpaceX under the microscope. For years, it’s faced suits for accidents and even deaths, including a botched rocket test that left one employee in a permanent coma when a piece of one of Starship’s engines flew off and fractured his skull. A Reuters investigation in 2023 found at least 600 cases of workplace injuries at SpaceX that went unreported.

The company is under intense pressure to perfect Starship, which NASA has selected to perform a lunar landing in its upcoming Artemis III mission — a role that’s now under some doubt.

The company also allegedly has a track record of retaliating against employees. It has faced several lawsuits over sexual harassment, including some that implicated CEO Musk. In a 2024 lawsuit, eight former employees accused the company of illegally firing them after raising concerns over sexual harassment.

It’s not the only Musk-owned venture facing similar legal allegations, either. His automaker Tesla has also been hit with numerous lawsuits over brutal working conditions and workplace accidents, joined by widespread accounts of Musk impulsively firing employees in fits of rage, and allegedly threatening to deport an engineer for raising a critical safety issue.

More on SpaceX: Elon Musk’s Starship Explosion Endangered Hundreds of Airline Passengers

The post SpaceX Hit With Back to Back Lawsuits From Workers Who Say They Were Brutally Injured on the Job appeared first on Futurism.

🔗 Sumber: futurism.com