MAROKO133 Breaking ai: OpenAI Is Suddenly in Trouble Hari Ini

📌 MAROKO133 Update ai: OpenAI Is Suddenly in Trouble Hari Ini

ChatGPT maker OpenAI was dealt a one-two punch on Tuesday, in a pair of separate blows that could undermine the firm’s dominance in the AI field.

First, as The Economist reports, Microsoft and Nvidia announced that they had signed a $350 billion deal with competitor Anthropic, “strategic partnerships” that almost doubled the latter company’s September valuation.

Then, Google came in with the kill by releasing Gemini 3, its latest and “most intelligent model” that impressed in early tests, especially when compared to OpenAI’s latest GPT-5.1 model, which debuted just last week.

In short, OpenAI’s grip looks shakier than ever — despite the firm’s meteoric rise to fame following the advent of ChatGPT almost exactly three years ago.

Complicating matters is a skittish market. Concerns over a growing AI bubble sparked a major tech selloff earlier this month, with analysts balking at an enormous and growing gulf between astronomical valuations and relatively tiny revenues.

While Nvidia’s posting of better-than-expected revenues on Wednesday reignited investor enthusiasm, nobody knows how the story will end.

“There has been a lot of talk about an AI bubble,” a boisterous Nvidia CEO Jensen Huang told investors. “From our vantage point, we see something very different.”

Now that Nvidia has signed up to hook Anthropic up with its AI hardware for the first time, and Google is making measurable progress with its latest Gemini 3 model, the pressure on OpenAI — once the de facto AI product, especially for curious amateurs — is mounting.

Of particular concern for OpenAI CEO Sam Altman: the usership gap is continuing to shrink. Google says its Gemini app has 650 million monthly active users, while Altman claimed last month that ChatGPT hit 800 million weekly users.

As The Economist points out, OpenAI’s plan appears to be doubling down by spending even more cash. It’s already planning to spend north of $1.4 trillion on data center buildouts over the next couple of years — and it’s still burning through billions of dollars each quarter.

That strategy could spread uncertainty from Wall Street to private markets, making it even harder to justify sky-high valuations in light of some steep competition putting a damper on potential revenue.

Nobody knows what all of the pressure on OpenAI will amount to. Some expect Anthropic — and particularly Google — to continue undermining the company’s dominance.

“We’re in a situation where — because of Google’s size and space and their first-mover advantage in search — Gemini could take market share and cause OpenAI and others to fall behind,” JonesTrading chief market strategist Mike O’Rourke told the New York Times.

Others are convinced that the gauntlet has already been thrown.

“OpenAI has basically squandered the technical lead it once had; Google has caught up,” noted AI skeptic Gary Marcus wrote in a recent blog post.

More on OpenAI: OpenAI Blocks Toymaker After Its AI Teddy Bear Is Caught Telling Children Terrible Things

The post OpenAI Is Suddenly in Trouble appeared first on Futurism.

🔗 Sumber: futurism.com

📌 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