PulseAugur
EN
LIVE 06:47:44

New 'Clade' phenomenon shows LLMs guess answers from broad evidence

By PulseAugur Editorial · [1 sources] ·

Researchers have identified a phenomenon in retrieval-augmented language models called "Clade" (Category-Level Answering from Deficient Evidence). This occurs when models retrieve passages that are relevant but provide information at a broader category level than the question requires, leading the model to guess the specific answer instead of abstaining or searching further. This guessing behavior was observed to be incorrect about half the time on biomedical questions from PrimeKG and also replicated on ecological species interactions. While ontologies can help map broader categories to specific answers, current methods struggle to effectively utilize them, with string matching only finding the correct family on the page for a small percentage of cases. AI

IMPACT Highlights a specific failure mode in retrieval-augmented models, suggesting a need for better ontology integration for accurate fine-grained answers.

RANK_REASON Research paper detailing a new phenomenon in language models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

New 'Clade' phenomenon shows LLMs guess answers from broad evidence

How we ranked this

Signal score
39 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Research paper detailing a new phenomenon in language models. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
paper, model release
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. arXiv cs.AI TIER_1 English(EN) · S M Rayeed, Charles V. Stewart, Yao Ma ·

    CLADE: Category-Level Answering from Deficient Evidence in Retrieval-Augmented Language Models

    arXiv:2611.14355v1 Announce Type: new Abstract: Retrieval-augmented systems often retrieve passages that are accurate and relevant but state the answer at a coarser level of granularity than the question requires. For example, a question may ask which specific disease a drug trea…

RELATED ENTITIES

RELATED TOPICS