Infoblock alpha
cs-cl/llm-evaluation IB-arxiv-2608.18554-automation-loses-augmentation-5-of-7 measured

Across seven economically grounded real-world tasks, the model that won the automation ranking lost the augmentation ranking on five of seven tasks.

Active Evidence state is recorded explicitly and may be revised by a later event.

Observed quantities

n_tasks 7 · tasks_where_automation_winner_loses_augmentation 5

01

Primary source

  1. arxiv · 2608.18554 · v1 · abstract
Title
CentaurBench: Benchmarking LLM Capabilities on Augmenting vs. Automating Real-World Work Tasks
Authors
Pattaraphon Kenny Wongchamcharoen, Kris Gulati, Min Min Fong, Abhishek Nagaraj
Version
v1
Check
Support quote found word for word in the arXiv abstract v1, 2026-10-08
Supporting quote from the source
the automation winner loses augmentation on five of seven tasks

02

Boundary of this block

No additional public limitation was recorded.

03

Connections

Outgoing

No outgoing public edges.

Incoming

No incoming public edges.