MAROKO133 Hot ai: Sugar-powered sight? 400-year eye mystery behind birds’ oxygen-free visi

📌 MAROKO133 Eksklusif ai: Sugar-powered sight? 400-year eye mystery behind birds’

Biologists have finally solved a centuries-old mystery regarding how birds maintain sharp eyesight despite their retinas lacking a direct blood supply. This means that bird retinas work without oxygen.

Aarhus University in Denmark has found that birds solve the problem of oxygen deprivation by switching their eyes to anaerobic power.

A bloodless retina that refuses to fail

Neural tissues, especially the retina, are among the most energy-demanding tissues in the body. 

These tissues typically require a constant supply of oxygen via blood vessels to function. However, bird retinas lack blood vessels, likely an evolutionary adaptation to prevent light scattering and improve visual clarity. 

Logically, these cells should be dead. Without blood, there is no oxygen. Without oxygen, neural tissue usually withers in minutes. 

“According to everything we know about physiology, this tissue should not be able to function,” said Christian Damsgaard, biologist and first author. 

Yet, birds see better than many other creatures on Earth.

Now, after eight years of investigation, the Aarhus team has finally decoded this vascular paradox.

But first, a quick rewind. Since the 1600s, biologists have pointed to the pecten oculi — a strange, comb-like structure protruding into the bird’s eye — as the secret oxygen tank. They assumed it leaked oxygen into the surrounding fluid to keep the retina alive.

It turns out, the experts’ assumptions were wrong. The researchers stated that the “pecten does not deliver oxygen to the retina at all.”

Instead, the tissue exists in a state of chronic deprivation that would cause a human brain to undergo a massive stroke.

Energy without oxygen

This raised a serious question: how does the retina stay powered? 

The team turned to “molecular GPS” called spatial transcriptomics, which enables mapping thousands of genes in the eye. 

It was found that oxygen-deprived inner layers have switched from oxygen-based metabolism to anaerobic glycolysis — a process that breaks down sugar in the absence of oxygen. 

This is a desperate, inefficient way to make energy. It produces 15 times less power per sugar molecule than oxygen-based breathing. 

“This mismatch raised yet another question: How can one of the most energy-hungry tissues in the body survive on such an inefficient process?” questioned Jens Randel Nyengaard, senior author. 

To solve the mystery, researchers used metabolic imaging to track sugar movement, discovering that bird retinas consume glucose at far higher rates than the rest of the brain. 

Here, the role of the pecten oculi — a structure long misunderstood — comes into play.

Pecten acts as a high-capacity “metabolic gateway.” The structure floods the retina with sugar to fuel its anaerobic engine and rapidly vacuums out lactate waste, preventing the build-up of toxic byproducts in the bloodless eye.

“The pecten is not an oxygen supplier. It is a transport system for fuel in and waste out,” said Nyengaard.

Understanding brain stroke

Evolutionary evidence suggests that birds inherited these bloodless retinas from their dinosaur ancestors to provide superior visual clarity.

The team says the findings could offer a potential medical blueprint for treating human strokes. 

In a brain stroke, human tissues die quickly when deprived of oxygen and choked by waste. Birds have evolved a natural strategy for surviving these exact conditions.

“Nature has solved a physiological problem in birds that makes humans sick,” Nyengaard added.

One day, it could help develop new strategies for protecting human brain tissue when blood flow is restricted or cut off.

The study was published in the journal Nature on January 21.

🔗 Sumber: interestingengineering.com

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

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