📌 TOPINDIATOURS Eksklusif ai: MIT Researchers Unveil “SEAL”: A New Step Towards Se

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

📌 TOPINDIATOURS Hot ai: Trump signs proclamation hitting tech industry with $100,0

U.S. President Donald Trump signed a proclamation on September 19, imposing a $100,000 annual fee on each H-1B visa application. The H-1B program has long been the backbone of the U.S. tech industry, especially for companies hiring skilled workers from India and China.

At a press briefing, US Commerce Secretary Howard Lutnick said the new rule was discussed with “all the big companies.” He added, “A hundred-thousand dollars a year for H-1B visas, and all of the big companies are on board. We’ve spoken to them.”

Lutnick stressed that firms should focus on hiring US graduates instead of relying on international talent. “Train Americans. Stop bringing in people to take our jobs,” he said.

Industry pushback and economic stakes

The announcement has triggered strong reactions. The tech industry, which has historically relied on H-1B workers more than any other sector, views the move as a major setback.

Amazon secured more than 10,000 H-1B approvals in the first half of 2025, while Microsoft and Meta each obtained more than 5,000.

Supporters of the visa system, like Tesla CEO Elon Musk, argue that it is vital for bringing in global talent. Musk himself once held an H-1B visa.

Critics, however, believe the program lowers wages for US workers and allows companies to bypass American talent.

Deedy Das, a partner at Menlo Ventures, warned that such high fees could hurt the economy. Writing on X, he said, “If the US ceases to attract the best talent, it drastically reduces its ability to innovate and grow.”

Legal questions and immigration policy shifts

Immigration experts are challenging the legality of Trump’s proclamation. Aaron Reichlin-Melnick, a senior fellow at the American Immigration Council, said on Bluesky: “The president has literally zero legal authority to impose a $100,000 fee on visas. None. Zip. Zilch.”

Currently, H-1B applicants pay a few thousand dollars in processing costs, most of which fall on employers. The program provides 65,000 visas each year, plus 20,000 more for those with advanced degrees. Approved visas allow workers to stay for three to six years.

The new fee is part of a larger set of actions by Trump aimed at tightening legal immigration. Last month, the US tested a program requiring bonds of up to $15,000 for certain visas. Earlier this year, a travel ban restricted entry from 19 countries. Previous attempts to limit H-1B visas were struck down by federal courts.

“Gold Card” for million-dollar residents

Alongside the H-1B announcement, Trump signed an executive order creating a new “gold card” program. This would allow individuals to gain US permanent residency if they can pay $1 million.

For many smaller firms and startups, the $100,000 annual H-1B fee could make hiring foreign talent impossible.

Larger corporations may absorb the cost, but experts warn that the overall impact could reduce America’s ability to compete for top global talent.

🔗 Sumber: interestingengineering.com