arXiv 2306.111672023-11-08cs.CLcs.AIcs.LG

大语言模型被红鲱鱼（误导性线索）固着：利用《Only Connect》连线墙数据集探索创造性问题解决与定势效应

Large Language Models are Fixated by Red Herrings: Exploring Creative Problem Solving and Einstellung Effect using the Only Connect Wall Dataset

Kite Research Institute, Toronto Rehabilitation Institute, University Health Network（Kite研究所，多伦多康复研究所，大学健康网络）；Department of Computer Science, University of Toronto（多伦多大学计算机科学系）；Institute of Biomedical Engineering, University of Toronto（多伦多大学生物医学工程研究所）；Vector Institute for AI（向量人工智能研究所）；Donnelly Centre for Cellular & Biomolecular Research, University of Toronto（多伦多大学唐纳利细胞与生物分子研究中心）；Department of Biomedical Data Science, Stanford University（斯坦福大学生物医学数据科学系）

Saeid Naeini, Raeid Saqur, Mozhgan Saeidi, John Giorgi, Babak Taati

备注：: v4,v3: Mincor cosmetic adjustments, typo-fixes etc. from V2. Fixed Fig. 2 caption overlapping with text in S2.2. V2: with added OCW-Randomized and OCW-WordNet results in Section 4.3 (added). 22 pages with Appendix

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本研究提出含刻意误导线索的OCW数据集及两个衍生数据集，评估大语言模型的创造性问题解决能力，探索其认知层面的固着与定势效应。

摘要：

自人工智能研究诞生以来，对类人人工智能的探索一直是该领域的长期课题。最新一批大语言模型（LLM）的技术演进与新兴能力，让这一主题跳出学术圈，成为大众文化热点。尽管近期的自然语言处理评估基准任务测试了类人行为的某些方面（例如BIG-bench的“类人行为”任务），但鲜有（即便有也极少）任务考察创造性问题解决能力。人类的创造性问题解决是认知神经科学中已被充分研究的课题，相关标准化测试主要以受试者建立线索词之间（异质）关联的能力作为创造力衡量指标。接触误导性刺激——即被称为“红鲱鱼”的干扰项——会通过固着效应和定势（Einstellung）范式降低人类在这类任务中的表现。在认知神经科学研究中，这类固着可通过让受试者预先接触与后续词片或线索正字法相似的错误词来实验性诱发。英国热门智力竞赛节目《Only Connect》的“连线墙”环节本质上模拟了梅德尼克远距离联想测验（RAT）的设计，内置了刻意设置的红鲱鱼，这使其成为探索和研究大语言模型中认知神经科学范畴的固着效应与定势范式的理想代理数据集。本文中，我们提出了全新的Only Connect Wall（OCW，《Only Connect》连线墙）数据集，并报告了对选定预训练语言模型和大语言模型的评估结果，评估的创造性问题解决任务包括按异质关联对线索词分组、识别各组中正确的开放知识领域关联。我们还合成生成了另外两个数据集：OCW-Randomized（OCW随机化版）、OCW-WordNet（OCW词网版），以进一步分析语言模型中的红鲱鱼假说。代码与数据集链接可在https://github.com/TaatiTeam/OCW获取。

英文摘要：

The quest for human imitative AI has been an enduring topic in AI research since its inception. The technical evolution and emerging capabilities of the latest cohort of large language models (LLMs) have reinvigorated the subject beyond academia to the cultural zeitgeist. While recent NLP evaluation benchmark tasks test some aspects of human-imitative behaviour (e.g., BIG-bench's 'human-like behavior' tasks), few, if not none, examine creative problem solving abilities. Creative problem solving in humans is a well-studied topic in cognitive neuroscience with standardized tests that predominantly use the ability to associate (heterogeneous) connections among clue words as a metric for creativity. Exposure to misleading stimuli - distractors dubbed red herrings - impede human performance in such tasks via the fixation effect and Einstellung paradigm. In cognitive neuroscience studies, such fixations are experimentally induced by pre-exposing participants to orthographically similar incorrect words to subsequent word-fragments or clues. The popular British quiz show Only Connect's Connecting Wall segment essentially mimics Mednick's Remote Associates Test (RAT) formulation with built-in, deliberate red herrings, which makes it an ideal proxy dataset to explore and study fixation effect and Einstellung paradigm from cognitive neuroscience in LLMs. In this paper we present the novel Only Connect Wall (OCW) dataset and report results from our evaluation of selected pre-trained language models and LLMs on creative problem solving tasks like grouping clue words by heterogeneous connections, and identifying correct open knowledge domain connections in respective groups. We synthetically generate two additional datasets: OCW-Randomized, OCW-WordNet to further analyze our red-herrings hypothesis in language models. The code and link to the dataset are available at https://github.com/TaatiTeam/OCW.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。