IAO Prompting: Making Knowledge Flow Explicit in LLMs through Structured Reasoning Templates

Aïssatou Diallo, Antonis Bikakis, Luke Dickens, Anthony Hunter, Rob Miller · 2025

Read IAO Prompting: Making Knowledge Flow Explicit in LLMs through Structured Reasoning Templates by Aïssatou Diallo, Antonis Bikakis, Luke Dickens, Anthony Hunter, Rob Miller (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-03080.txt.

Description

Aïssatou Diallo and four colleagues make language models show their knowledge trail with a structured prompting scheme, in a 2025 paper. Chain-of-thought prompting exposes reasoning steps, but which facts the model used, and where it might have gone wrong, stay hidden. Their method, IAO prompting, attacks that opacity.

Every step of the solution must name its Input knowledge, the Action performed, and the Output produced. The explicit flow lets users trace how stored knowledge moves through a problem. Factual consistency becomes checkable, and gaps or misapplied facts stand out. Experiments across varied reasoning tasks showed better zero-shot performance than standard prompting. Human judges confirmed the traces were easier to verify, and better at exposing hallucinations and reasoning slips. The authors pitch the template as both a performance trick and an audit tool for high-stakes model use.

Catalog record

CREATORAïssatou Diallo, Antonis Bikakis, Luke Dickens, Anthony Hunter, Rob Miller
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE75 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-03080

Browse the full library · About the Tower of Babel