When Slower Isn't Truer: Inverse Scaling Law of Truthfulness in Multimodal Reasoning

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Fang, Sitong, Cao, Wenjing, Li, Jiahao, Wang, Xuyao, Dai, Juntao, Chan, Chi-Min, Han, Sirui, Guo, Yike, Yang, Yaodong, Ji, Jiaming
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911596891078656
author Fang, Sitong
Cao, Wenjing
Li, Jiahao
Wang, Xuyao
Dai, Juntao
Chan, Chi-Min
Han, Sirui
Guo, Yike
Yang, Yaodong
Ji, Jiaming
author_facet Fang, Sitong
Cao, Wenjing
Li, Jiahao
Wang, Xuyao
Dai, Juntao
Chan, Chi-Min
Han, Sirui
Guo, Yike
Yang, Yaodong
Ji, Jiaming
contents Reasoning models have attracted increasing attention for their ability to tackle complex tasks, embodying the System II (slow thinking) paradigm in contrast to System I (fast, intuitive responses). Yet a key question remains: Does slower reasoning necessarily lead to more truthful answers? Our findings suggest otherwise. We conduct the first systematic study of the inverse scaling law in slow-thinking paradigms for multimodal reasoning. We find that when confronted with incomplete or misleading visual inputs, slow-thinking models are more prone to fabricating plausible yet false details to justify untruthful reasoning. To analyze this behavior, we construct a 5,000-sample hierarchical prompt dataset annotated by 50 human participants. The prompts progressively increase in complexity, revealing a consistent pattern: slower reasoning models tend to follow depth-first search (DFS) thinking, persistently exploring flawed premises, while faster chat models favor breadth-first search (BFS) inference, showing greater caution under uncertainty. These findings reveal a critical vulnerability of reasoning models: while effective in structured domains such as math, their DFS-style reasoning becomes fragile when confronted with ambiguous, multimodal inputs.
format Preprint
id arxiv_https___arxiv_org_abs_2505_20214
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle When Slower Isn't Truer: Inverse Scaling Law of Truthfulness in Multimodal Reasoning
Fang, Sitong
Cao, Wenjing
Li, Jiahao
Wang, Xuyao
Dai, Juntao
Chan, Chi-Min
Han, Sirui
Guo, Yike
Yang, Yaodong
Ji, Jiaming
Artificial Intelligence
Reasoning models have attracted increasing attention for their ability to tackle complex tasks, embodying the System II (slow thinking) paradigm in contrast to System I (fast, intuitive responses). Yet a key question remains: Does slower reasoning necessarily lead to more truthful answers? Our findings suggest otherwise. We conduct the first systematic study of the inverse scaling law in slow-thinking paradigms for multimodal reasoning. We find that when confronted with incomplete or misleading visual inputs, slow-thinking models are more prone to fabricating plausible yet false details to justify untruthful reasoning. To analyze this behavior, we construct a 5,000-sample hierarchical prompt dataset annotated by 50 human participants. The prompts progressively increase in complexity, revealing a consistent pattern: slower reasoning models tend to follow depth-first search (DFS) thinking, persistently exploring flawed premises, while faster chat models favor breadth-first search (BFS) inference, showing greater caution under uncertainty. These findings reveal a critical vulnerability of reasoning models: while effective in structured domains such as math, their DFS-style reasoning becomes fragile when confronted with ambiguous, multimodal inputs.
title When Slower Isn't Truer: Inverse Scaling Law of Truthfulness in Multimodal Reasoning
topic Artificial Intelligence
url https://arxiv.org/abs/2505.20214