📌 TOPINDIATOURS Update ai: Researchers from PSU and Duke introduce “Multi-Agent Sy

Share My Research is Synced’s column that welcomes scholars to share their own research breakthroughs with over 2M global AI enthusiasts. Beyond technological advances, Share My Research also calls for interesting stories behind the research and exciting research ideas. 

Meet the author
Institutions: Penn State University, Duke University, Google DeepMind, University of Washington, Meta, Nanyang Technological University, and Oregon State University. The co-first authors are Shaokun Zhang of Penn State University and Ming Yin of Duke University.

In recent years, LLM Multi-Agent systems have garnered widespread attention for their collaborative approach to solving complex problems. However, it’s a common scenario for these systems to fail at a task despite a flurry of activity. This leaves developers with a critical question: which agent, at what point, was responsible for the failure? Sifting through vast interaction logs to pinpoint the root cause feels like finding a needle in a haystack—a time-consuming and labor-intensive effort.
 
This is a familiar frustration for developers. In increasingly complex Multi-Agent systems, failures are not only common but also incredibly difficult to diagnose due to the autonomous nature of agent collaboration and long information chains. Without a way to quickly identify the source of a failure, system iteration and optimization grind to a halt.
 
To address this challenge, researchers from Penn State University and Duke University, in collaboration with institutions including Google DeepMind, have introduced the novel research problem of “Automated Failure Attribution.” They have constructed the first benchmark dataset for this task, Who&When, and have developed and evaluated several automated attribution methods. This work not only highlights the complexity of the task but also paves a new path toward enhancing the reliability of LLM Multi-Agent systems.
The paper has been accepted as a Spotlight presentation at the top-tier machine learning conference, ICML 2025, and the code and dataset are now fully open-source.

Paper：https://arxiv.org/pdf/2505.00212
Code：https://github.com/mingyin1/Agents_Failure_Attribution
Dataset：https://huggingface.co/datasets/Kevin355/Who_and_When
 
 
Research Background and Challenges
LLM-driven Multi-Agent systems have demonstrated immense potential across many domains. However, these systems are fragile; errors by a single agent, misunderstandings between agents, or mistakes in information transmission can lead to the failure of the entire task.

Currently, when a system fails, developers are often left with manual and inefficient methods for debugging:
Manual Log Archaeology : Developers must manually review lengthy interaction logs to find the source of the problem.
Reliance on Expertise : The debugging process is highly dependent on the developer’s deep understanding of the system and the task at hand.
 
This “needle in a haystack” approach to debugging is not only inefficient but also severely hinders rapid system iteration and the improvement of system reliability. There is an urgent need for an automated, systematic method to pinpoint the cause of failures, effectively bridging the gap between “evaluation results” and “system improvement.”

Core Contributions
This paper makes several groundbreaking contributions to address the challenges above:
1. Defining a New Problem: The paper is the first to formalize “automated failure attribution” as a specific research task. This task is defined by identifying the

2. failure-responsible agent and the decisive error step that led to the task’s failure.

Constructing the First Benchmark Dataset: Who&When : This dataset includes a wide range of failure logs collected from 127 LLM Multi-Agent systems, which were either algorithmically generated or hand-crafted by experts to ensure realism and diversity. Each failure log is accompanied by fine-grained human annotations for:
Who: The agent responsible for the failure.
When: The specific interaction step where the decisive error occurred.
Why: A natural language explanation of the cause of the failure.

3. Exploring Initial “Automated Attribution” Methods : Using the Who&When dataset, the paper designs and assesses three distinct methods for automated failure attribution:
All-at-Once: This method provides the LLM with the user query and the complete failure log, asking it to identify the responsible agent and the decisive error step in a single pass. While cost-effective, it may struggle to pinpoint precise errors in long contexts.
Step-by-Step: This approach mimics manual debugging by having the LLM review the interaction log sequentially, making a judgment at each step until the error is found. It is more precise at locating the error step but incurs higher costs and risks accumulating errors.
Binary Search: A compromise between the first two methods, this strategy repeatedly divides the log in half, using the LLM to determine which segment contains the error. It then recursively searches the identified segment, offering a balance of cost and performance.
 
Experimental Results and Key Findings

Experiments were conducted in two settings: one where the LLM knows the ground truth answer to the problem the Multi-Agent system is trying to solve (With Ground Truth) and one where it does not (Without Ground Truth). The primary model used was GPT-4o, though other models were also tested. The systematic evaluation of these methods on the Who&When dataset yielded several important insights:

• A Long Way to Go: Current methods are far from perfect. Even the best-performing single method achieved an accuracy of only about 53.5% in identifying the responsible agent and a mere 14.2% in pinpointing the exact error step. Some methods performed even worse than random guessing, underscoring the difficulty of the task.
• No “All-in-One” Solution: Different methods excel at different aspects of the problem. The All-at-Once method is better at identifying “Who,” while the Step-by-Step method is more effective at determining “When.” The Binary Search method provides a middle-ground performance.

• Hybrid Approaches Show Promise but at a High Cost: The researchers found that combining different methods, such as using the All-at-Once approach to identify a potential agent and then applying the Step-by-Step method to find the error, can improve overall performance. However, this comes with a significant increase in computational cost.

• State-of-the-Art Models Struggle: Surprisingly, even the most advanced reasoning m…

  Konten dipersingkat otomatis.

  🔗 Sumber: syncedreview.com

  ---

  📌 TOPINDIATOURS Eksklusif ai: Bernie Sanders Calls for Breakup of OpenAI Wajib Bac

  Vermont Senator Bernie Sanders is publicly calling on the government to break up ChatGPT maker OpenAI.

  When Axios asked him whether the Sam Altman-led company should be broken up, he responded with “I do.”

  “We need to take a deep breath and understand it’s like a meteor coming,” he told the publication in an interview last week. “We’ve got to be prepared to deal with all of its complexity.”

  To be clear, the idea is a serious long shot. While president Donald Trump ran on the promise of breaking up big tech, major companies spent millions on currying favor with him — including donations for his latest White House ballroom project — with varying levels of success.

  Trump has also embraced AI fullheartedly during his second term, proudly announcing a massive $500 billion AI infrastructure project, dubbed Stargate, soon after taking office. The initiative involves OpenAI, among other entities.

  Whether OpenAI is an anti-competitive monopoly in need of breaking up remains a point of contention. Plenty of alternatives to its AI models exist, including Google’s Gemini and Anthropic’s Claude.

  In a statement to Axios, OpenAI’s head of policy communications, Liz Bourgeois, argued that the company is “building in a field shaped for decades by a few large technology companies with deep resources and structural advantages.”

  “Our growth reflects something simple: People find what we’re building useful,” she added. “This is what healthy competition looks like in the US — offering better choices.”

  Others have called for regulatory intervention following concerns of a growing AI bubble, which could leave the US economy in ruins in case it were to burst.

  “The first step to a healthier market is to break up companies that are vertically integrated, so that platforms don’t compete with their customers,” Vanderbilt Policy Accelerator director of AI and technology policy Asad Ramzanali wrote in a recent piece for TIME.

  Ramzanali pointed at AI chipmaker Nvidia’s eyebrow-raising $100 billion investment in OpenAI, a “circular” deal that fueled fears over a looming AI bubble.

  “Chips must be independent from clouds, and clouds must be independent from AI models,” he wrote. “Those models should compete on merit, not on whether they’re tethered to a trillion-dollar sponsor.”

  Besides calling for the breakup of OpenAI, Sanders argued during last week’s interview that AI could have an “enormous transformational impact” while also potentially causing “massive” job losses.

  “I want to see small businesses develop,” he added. “I want to see creativity out there in the economy. Ain’t going to do any good for the younger people if the entry level jobs are taken over by AI.”

  The senator also said he’s concerned about “how we relate to each other as human beings,” while taking direct aim at AI startup Friend, which made headlines for its controversial wearable that’s designed to act as a companion. The company’s New York subway ad campaign drew major backlash and ignited a heated discussion surrounding the role of AI in our daily lives.

  The news comes after Sanders shared a report earlier this month about the impact AI could have on jobs over the next decade.

  “AI and automation could destroy nearly 100 million US jobs in a decade,” the report reads, noting that the “economic gains” in AI “have gone almost exclusively to those at the top.”

  In response, Sanders called for a “robot tax” to be levied against large corporations to distribute to workers whose lives are upended by technological automation, a radical solution that will likely draw plenty of skepticism. For one, the vast majority of companies are currently struggling to generate revenue using AI.

  In short, Sanders is making no secret about his disdain for the AI industry’s disregard for working-class people.

  In a separate interview with Vanity Fair last week, Bernie took aim at Trump’s personal embrace of AI.

  “Trump is taking this country in a whole new direction,” he said. “This is a guy who puts up an AI image of him in an airplane defecating on American cities — not quite the image of the president of the United States that I was educated to respect when I was in the fourth grade,” Sanders added, referring to a controversial AI-generated video Trump posted on social media, depicting him releasing feces on protesters.

  More on Bernie Sanders: Bernie Sanders Has a Fascinating Idea About How to Prevent AI From Wiping Out the Economy

  The post Bernie Sanders Calls for Breakup of OpenAI appeared first on Futurism.

  🔗 Sumber: futurism.com

  ---