Making AI risk decisions visible: evaluating LLM reasoning in insurance risk classification

Andrea Rosales Sanabria

doi:10.1145/3742414.3794728

Making AI risk decisions visible: evaluating LLM reasoning in insurance risk classification

Andrea Rosales Sanabria^*

^*Corresponding author for this work

School of Computer Science

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

In the insurance industry, accurate risk assessment is a critical part of underwriting. With increasing attention on exploring LLM in assist risk assessment, we are investigating how to quantitatively measure the reasoning quality of LLM. We develop a user interface in Streamit that visualises and presents these measurements to support human underwriters in informed decision-making.

Original language	English
Title of host publication	Companion proceedings of the 31st international conference on intelligent user interfaces (IUI '26)
Place of Publication	New York, NY
Publisher	ACM
Pages	81-84
Number of pages	4
ISBN (Print)	9798400719851
DOIs	https://doi.org/10.1145/3742414.3794728
Publication status	Published - 22 Mar 2026
Event	31st Annual ACM Conference on Intelligent User Interfaces - St Raphael Resort & Marin, Limassol, Cyprus Duration: 13 Jul 2026 → 16 Jul 2026 https://iui.acm.org/2026/

Conference

Conference	31st Annual ACM Conference on Intelligent User Interfaces
Abbreviated title	IUI 2026
Country/Territory	Cyprus
City	Limassol
Period	13/07/26 → 16/07/26
Internet address	https://iui.acm.org/2026/

Access to Document

10.1145/3742414.3794728Licence: CC BY

Rosales-Sanabria_2026_IUI_Making-AI-risk-decisions-visible_CC
© 2026 Copyright held by the owner/author(s). This work is licensed under a Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0).
Final published version, 960 KBLicence: CC BY

Handle.net

Persistent link to portal record

Cite this

@inproceedings{04dca0067248436a90794bdf7b85ae9b,

title = "Making AI risk decisions visible: evaluating LLM reasoning in insurance risk classification",

abstract = "In the insurance industry, accurate risk assessment is a critical part of underwriting. With increasing attention on exploring LLM in assist risk assessment, we are investigating how to quantitatively measure the reasoning quality of LLM. We develop a user interface in Streamit that visualises and presents these measurements to support human underwriters in informed decision-making.",

author = "\{Rosales Sanabria\}, Andrea",

note = "Funding: This work is funded by InnovateUK 10094941.; 31st Annual ACM Conference on Intelligent User Interfaces, IUI 2026 ; Conference date: 13-07-2026 Through 16-07-2026",

year = "2026",

month = mar,

day = "22",

doi = "10.1145/3742414.3794728",

language = "English",

isbn = "9798400719851",

pages = "81--84",

booktitle = "Companion proceedings of the 31st international conference on intelligent user interfaces (IUI '26)",

publisher = "ACM",

address = "United States",

url = "https://iui.acm.org/2026/",

}

Rosales Sanabria, A 2026, Making AI risk decisions visible: evaluating LLM reasoning in insurance risk classification. in Companion proceedings of the 31st international conference on intelligent user interfaces (IUI '26). ACM, New York, NY, pp. 81-84, 31st Annual ACM Conference on Intelligent User Interfaces, Limassol, Cyprus, 13/07/26. https://doi.org/10.1145/3742414.3794728

TY - GEN

T1 - Making AI risk decisions visible

T2 - 31st Annual ACM Conference on Intelligent User Interfaces

AU - Rosales Sanabria, Andrea

N1 - Funding: This work is funded by InnovateUK 10094941.

PY - 2026/3/22

Y1 - 2026/3/22

N2 - In the insurance industry, accurate risk assessment is a critical part of underwriting. With increasing attention on exploring LLM in assist risk assessment, we are investigating how to quantitatively measure the reasoning quality of LLM. We develop a user interface in Streamit that visualises and presents these measurements to support human underwriters in informed decision-making.

AB - In the insurance industry, accurate risk assessment is a critical part of underwriting. With increasing attention on exploring LLM in assist risk assessment, we are investigating how to quantitatively measure the reasoning quality of LLM. We develop a user interface in Streamit that visualises and presents these measurements to support human underwriters in informed decision-making.

UR - http://doi.org/10.1145/3742414

U2 - 10.1145/3742414.3794728

DO - 10.1145/3742414.3794728

M3 - Conference contribution

SN - 9798400719851

SP - 81

EP - 84

BT - Companion proceedings of the 31st international conference on intelligent user interfaces (IUI '26)

PB - ACM

CY - New York, NY

Y2 - 13 July 2026 through 16 July 2026

ER -

Making AI risk decisions visible: evaluating LLM reasoning in insurance risk classification

Abstract

Conference

Access to Document

Handle.net

Other files and links

Fingerprint

Cite this