MAROKO133 Eksklusif ai: Amazon’s robotaxis make risky intersection stops, prompting 332-ve

📌 MAROKO133 Breaking ai: Amazon’s robotaxis make risky intersection stops, prompti

Amazon-owned autonomous vehicle company Zoox has announced a voluntary software recall after identifying driving behavior near intersections that could raise safety concerns, according to federal filings released Tuesday.

The recall applies to 332 driverless vehicles and involves software used by Zoox robotaxis operating on public roads.

The National Highway Traffic Safety Administration said the affected vehicles may cross yellow center lines, block crosswalks, or stop in front of oncoming traffic near intersections.

Zoox reported no crashes tied to the issue but acknowledged an increased risk.

Zoox currently runs public robotaxi services in parts of San Francisco and Las Vegas, where it offers free rides in fully autonomous vehicles.

Turning behavior reviewed

Zoox first detected the issue in late August after a robotaxi executed a wide right turn near an intersection. According to NHTSA documents, the vehicle crossed partially into the opposing lane and paused in front of oncoming traffic.

That incident prompted a broader internal review. Zoox analyzed driving data and identified 62 cases between August 26 and December 5 where vehicles crossed lane markings unnecessarily near intersections.

Some crossings were partial, while others extended fully into opposing lanes.

Zoox told regulators it remained engaged with federal officials throughout the review.

The company said it was in “ongoing conversations with NHTSA about the frequency, severity, and root causes of these occurrences.”

A Zoox spokesperson said the company identified driving actions that did not align with its internal safety standards.

In certain cases, robotaxis stopped inside crosswalks to avoid blocking intersections at red lights. In other situations, vehicles completed turns too late, resulting in wide maneuvers.

Software fixes applied

Zoox said it resolved the problem through software updates issued on November 7 and again in mid-December.

The recall documents those updates rather than requiring physical vehicle repairs.

“We have successfully identified and deployed targeted software improvements to address the root causes of these incidents,” the company said.

“Today, we’re submitting a voluntary software recall because transparency and safety is foundational to Zoox, and we want to be open with the public and regulators about how we are constantly refining and improving our technology.”

The recall covers Zoox vehicles that operated on public roads between March 13 and December 18.

The company said the updated software prevents the behaviors identified during the review.

The latest recall adds to a growing list of software fixes Zoox has issued this year.

In March, the company recalled vehicles after reports of unexpected hard braking.

That action followed two incidents where motorcyclists struck the rear of Zoox vehicles.

Zoox also issued recalls in May to improve how its system predicts the movement of pedestrians and other road users.

One update followed an April crash involving an unoccupied robotaxi and a passenger vehicle in Las Vegas.

Federal regulators have recently closed several probes involving Zoox.

The NHTSA ended a braking investigation in April and certified Zoox vehicles for demonstration use in August, closing a separate compliance probe that began in 2022.

Other autonomous vehicle developers face similar scrutiny.

Alphabet-owned Waymo recently recalled vehicles after Texas officials reported illegal school bus passings.

The NHTSA opened an investigation into that matter in October.

As autonomous services expand, regulators continue to monitor software performance closely, especially in complex urban traffic environments.

🔗 Sumber: interestingengineering.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