生成式AI回答事實(shí)型參考咨詢問(wèn)題的效果與啟示

打開(kāi)文本圖片集

中圖分類(lèi)號(hào)：G252.61；TP18 文獻(xiàn)標(biāo)識(shí)碼：A

Abstract As large language model technology continues to mature，generative AI is reshaping various industries.This studyevaluatestheefectivenessofgenerativeAI inrespondingtofact-basedreferencequestions withinreal-worldlibrary service scenarios.Teselectedquestions，characterizedbystrongobjectivitycover fourcategories：statisticalinformation， noun definitions，origins of allusions，and historicalresearch.Four generative AI models werechosen to generate answers based onthe original queries，and their responses were compared with standard replies providedby librarians to assess performance.The analysis reveals that while generative AI demonstrates strong capabilities in informationretrieval and generation，its performanceinprovidingauthoritative sources remains inconsistent.Moreover，itlackstheability to deeplyunderstand and integrate information to answer more specialized inquiries.The future of reference services necesitates that libraryprofessionals leverage theirinnovativecapacities tocontinuously explore and refine strateges for AI integration.

KeyWordsartificialintellgence;referencesvices;largelanguagemodel;smartlibrary;artificialintelligencegeated content

1引言

2022年11月，OpenAI的生成式人工智能（ArtificialIntelligence，AI）應(yīng)用ChatGPT上線，其智能化解析自然語(yǔ)言輸入并生成連貫文本的能力為問(wèn)答系統(tǒng)以及人工智能生成內(nèi)容帶來(lái)了顛覆性變革。（剩余5270字）