Robust knowledge extraction from large language models using social choice theory
Robust knowledge extraction from large language models using social choice theory
Large-language models (LLMs) have the potential to support a wide range of applications like conversational agents, creative writing, text improvement, and general query answering. However, they are ill-suited for query answering in high-stake domains like medicine because they generate answers at random and their answers are typically not robust - even the same query can result in different answers when prompted multiple times. In order to improve the robustness of LLM queries, we propose using ranking queries repeatedly and to aggregate the queries using methods from social choice theory. We study ranking queries in diagnostic settings like medical and fault diagnosis and discuss how the Partial Borda Choice function from the literature can be applied to merge multiple query results. We discuss some additional interesting properties in our setting and evaluate the robustness of our approach empirically.
Potyka, Nico
a8a29aeb-d747-4ac0-9c76-b093b4d3bb67
Zhu, Yuqicheng
e2164ad8-3ba3-4dd5-9a6f-81f253f938c6
He, Yunjie
e92ec3e3-2008-464b-9e37-694ad05264aa
Kharlamov, Evgenyi
5a522384-6a70-4f2c-ab5a-6b95c1944b32
Staab, Steffen
bf48d51b-bd11-4d58-8e1c-4e6e03b30c49
Potyka, Nico
a8a29aeb-d747-4ac0-9c76-b093b4d3bb67
Zhu, Yuqicheng
e2164ad8-3ba3-4dd5-9a6f-81f253f938c6
He, Yunjie
e92ec3e3-2008-464b-9e37-694ad05264aa
Kharlamov, Evgenyi
5a522384-6a70-4f2c-ab5a-6b95c1944b32
Staab, Steffen
bf48d51b-bd11-4d58-8e1c-4e6e03b30c49
Potyka, Nico, Zhu, Yuqicheng, He, Yunjie, Kharlamov, Evgenyi and Staab, Steffen
(2023)
Robust knowledge extraction from large language models using social choice theory.
The 23rd International Conference on Autonomous Agents and Multi-Agent Systems, Cordis Hotel, Auckland, New Zealand.
06 - 10 May 2024.
14 pp
.
(In Press)
(doi:10.48550/arXiv.2312.14877).
Record type:
Conference or Workshop Item
(Paper)
Abstract
Large-language models (LLMs) have the potential to support a wide range of applications like conversational agents, creative writing, text improvement, and general query answering. However, they are ill-suited for query answering in high-stake domains like medicine because they generate answers at random and their answers are typically not robust - even the same query can result in different answers when prompted multiple times. In order to improve the robustness of LLM queries, we propose using ranking queries repeatedly and to aggregate the queries using methods from social choice theory. We study ranking queries in diagnostic settings like medical and fault diagnosis and discuss how the Partial Borda Choice function from the literature can be applied to merge multiple query results. We discuss some additional interesting properties in our setting and evaluate the robustness of our approach empirically.
Text
2312.14877
- Accepted Manuscript
More information
Accepted/In Press date: 20 December 2023
Venue - Dates:
The 23rd International Conference on Autonomous Agents and Multi-Agent Systems, Cordis Hotel, Auckland, New Zealand, 2024-05-06 - 2024-05-10
Identifiers
Local EPrints ID: 485892
URI: http://eprints.soton.ac.uk/id/eprint/485892
PURE UUID: 9cdc4d96-270e-4a9f-bf44-9c7c2cdf4450
Catalogue record
Date deposited: 03 Jan 2024 20:19
Last modified: 18 Mar 2024 03:32
Export record
Altmetrics
Contributors
Author:
Nico Potyka
Author:
Yuqicheng Zhu
Author:
Yunjie He
Author:
Evgenyi Kharlamov
Author:
Steffen Staab
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics