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

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

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: The Chinese Streaming Industry Is Being Gutted by AI-Gen

Earlier this year, TikTok’s Chinese owner ByteDance released the latest version of its Seedance AI video generating tool.

Impressively photorealistic footage of Will Smith battling a ferocious spaghetti monster or Brad Pitt and Tom Cruise engaging in hand-to-hand combat sent Hollywood into a frenzy, highlighting ongoing concerns over the status of human creativity in the age of AI.

It’s not just Hollywood struggling to adapt to a new reality. As the New York Times reports, Chinese directors, actors, and crew share these concerns. They’ve watched as generative AI has caused the number of “microdramas” — ultra-short-form serialized clips optimized for mobile viewing — being produced to skyrocket. The format has caught on like wildfire in China, quickly turning into a massive multibillion-dollar business.

According to the paper, some 50,000 new AI-generated microdramas were added to Douyin, China’s TikTok in March alone. Many of them are racking up hundreds of millions of views, in a growing AI-based content factory estimated to be worth more than $3 billion this year. (The microdrama industry overall is expected to exceed $16.5 billion by the end of the year.)

In other words, it’s no wonder that those who make their living in the country’s entertainment industry are crying foul, especially as controversies over the unlicensed use of the likenesses of Chinese entertainers continue to swirl. (Cases about replacing employees with AI are also currently making waves in the Chinese court system.)

Actor Li Jiao told the NYT that he watched as the number of available roles dried up. He suggested that the hype around AI may be at least partially to blame.

“It’s like it was raining, and then suddenly the rain stopped,” he said.

Microdrama director Wang Yushun admitted to the newspaper that he was making extensive use of AI, lamenting that he had to lay off employees and citing waning demand for live-action productions.

Meanwhile, competition in the industry is ramping up as AI massively lowers the barriers to entry.

The Chinese government is seemingly still on the back foot, with the country’s cyberspace regulator most recently issuing rules to require clear labeling of and consent for the creation of AI-generated “digital humans” while banning services that could get children addicted or misled.

Despite the threat to his career, Li says it’s not necessarily a matter of abandoning AI altogether, highlighting a far more nuanced embrace of the tech in China compared to the outright opposition among a growing cohort of Hollywood AI-listers.

“They’re still just imitating humans or trying to make things more humanlike,” he told the NYT. “They should be trying to unleash more imagination, taking a more unconventional route.”

“After all, our fundamental value as humans is in our ability to imagine,” he added.

More on China and AI: New AI Video Generator Is So Impressive That It’s Scaring Hollywood

The post The Chinese Streaming Industry Is Being Gutted by AI-Generated Shows appeared first on Futurism.

🔗 Sumber: futurism.com