怎么样经过主动式对话AI助手优化用户体验评定？——探索意见机会对分析效果的影响
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/sz_mmbiz_png/IJTyfegB2HtvB7WHezrtFLv6Mfib7Kgdo5d0rV4g7ibNPFAgPkU74l6jf8djtP9I2jz2Eg4h0nwkF3qTAoHheS2w/640?wx_fmt=png&from=appmsg&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">导语：</span><span style="color: black;">在人工智能技术日益渗透到各个<span style="color: black;">行业</span>的今天，其在用户体验（UX）<span style="color: black;">评定</span>中的应用<span style="color: black;">亦</span>变得愈发重要。<span style="color: black;">那样</span><span style="color: black;">怎样</span><span style="color: black;">经过</span>主动式对话AI助手来优化用户体验<span style="color: black;">评定</span>呢？今天给<span style="color: black;">大众</span>带来</span><span style="color: black;">Emily Kuang</span><span style="color: black;">等人<span style="color: black;">发布</span>于CHI上的一篇<span style="color: black;">文案</span>《Enhancing UX Evaluation Through Collaboration with Conversational AI Assistants: Effects of Proactive Dialogue and Timing》。欢迎<span style="color: black;">大众</span>跟随<span style="color: black;">博主</span>的步伐，探索AI<span style="color: black;">意见</span><span style="color: black;">机会</span>与方式对用户体验分析的影响。</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">快</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">速</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">阅</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">读</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">版</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/sz_mmbiz_png/IJTyfegB2HuHvO3pslKWTIibjTibjOXaoh6JOgKtsPX1mNBBdBw5uu2FEmAqsF6BWsYib8l2R03FCBDsjZstc25LQ/640?wx_fmt=png&from=appmsg&&wxfrom=5&wx_lazy=1&wx_co=1&tp=webp" style="width: 50%; margin-bottom: 20px;"></p><strong style="color: blue;"><span style="color: black;"><span style="color: black;">科研</span>背景：</span></strong>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">传统的可用性测试是<span style="color: black;">发掘</span>交互系统中可用性问题的重要<span style="color: black;">办法</span>，但分析测试视频既<span style="color: black;">繁杂</span>又资源密集。随着人工智能（AI）技术的进步，<span style="color: black;">科研</span>人员<span style="color: black;">起始</span>探索人类与AI合作进行UX分析的可能性，尤其是<span style="color: black;">经过</span>自然语言的方式。</span></p><strong style="color: blue;"><span style="color: black;"><span style="color: black;">科研</span>目的：</span></strong><span style="color: black;">探索主动式会话人工智能助手在<span style="color: black;">区别</span>时间点<span style="color: black;">供给</span>自动<span style="color: black;">意见</span><span style="color: black;">针对</span>用户体验（UX）<span style="color: black;">评定</span>者分析可用性问题的影响，<span style="color: black;">尤其</span>是比较这些<span style="color: black;">意见</span>在潜在问题<span style="color: black;">出现</span>之前、同步时和之后<span style="color: black;">显现</span>时的效果，并<span style="color: black;">评定</span>这种合作方式对分析性能、用户信任和效率的影响，以及<span style="color: black;">怎样</span>优化AI辅助工具的设计和功能，以<span style="color: black;">加强</span><span style="color: black;">将来</span>在UX<span style="color: black;">评定</span>中的人类-AI协作。</span><strong style="color: blue;"><span style="color: black;"><span style="color: black;">科研</span>过程：</span></strong><span style="color: black;"><span style="color: black;">经过</span>一项混合型Wizard-of-Oz<span style="color: black;">科研</span>，招募了24名UX<span style="color: black;">评定</span>者，<span style="color: black;">运用</span>ChatGPT<span style="color: black;">按照</span>可用性测试视频的转录文本生成自动问题<span style="color: black;">意见</span>，并<span style="color: black;">经过</span>真人演员（即wizard）<span style="color: black;">回复</span><span style="color: black;">评定</span>者即兴提出的问题。<span style="color: black;">科研</span>设计了三种<span style="color: black;">区别</span>的<span style="color: black;">意见</span><span style="color: black;">机会</span><span style="color: black;">要求</span>——在潜在问题之前、同步时和之后——以<span style="color: black;">评定</span>这些<span style="color: black;">机会</span>对<span style="color: black;">评定</span>者分析<span style="color: black;">行径</span>的影响。参与者在分析视频时，会<span style="color: black;">按照</span>预设的时间点接收到自动<span style="color: black;">意见</span>，并<span style="color: black;">经过</span>聊天窗口与AI助手互动，表达<span style="color: black;">她们</span>对<span style="color: black;">意见</span>的同意、纠正、澄清请求或<span style="color: black;">区别</span>意。<span style="color: black;">科研</span><span style="color: black;">经过</span>定量和定性数据分析，探讨了自动<span style="color: black;">意见</span>的<span style="color: black;">机会</span>对UX<span style="color: black;">评定</span>者分析性能和主观感知的影响，以及<span style="color: black;">她们</span>对AI生成<span style="color: black;">意见</span>的反应和接受程度。</span>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><strong style="color: blue;"><span style="color: black;"><span style="color: black;">科研</span>结果：</span></strong></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">自动<span style="color: black;">意见</span>的<span style="color: black;">机会</span>对UX<span style="color: black;">评定</span>者识别问题的总数<span style="color: black;">无</span><span style="color: black;">明显</span>影响，但大<span style="color: black;">都数</span>参与者更倾向于在潜在问题<span style="color: black;">出现</span>后接收<span style="color: black;">意见</span>，这种<span style="color: black;">机会</span>的安排<span style="color: black;">明显</span><span style="color: black;">加强</span>了<span style="color: black;">她们</span>的信任感和效率。尽管参与者认为ChatGPT生成的自动<span style="color: black;">意见</span>有用，但<span style="color: black;">她们</span>自己<span style="color: black;">发掘</span>的问题数量是AI的两倍以上，这<span style="color: black;">显示</span>人类专业知识在UX<span style="color: black;">评定</span>中仍然不可替代。<span style="color: black;">另外</span>，参与者对自动<span style="color: black;">意见</span>的响应多种多样，<span style="color: black;">包含</span>同意、纠正、请求澄清和<span style="color: black;">区别</span>意，其中77.6%的<span style="color: black;">意见</span>被接受。<span style="color: black;">科研</span>还<span style="color: black;">发掘</span>，尽管ChatGPT能够识别出<span style="color: black;">有些</span><span style="color: black;">关联</span>的可用性问题，但它<span style="color: black;">错失</span>了58.8%的由参与者集体识别的独特问题，这强调了在UX<span style="color: black;">评定</span>中结合人类专业知识和AI工具的重要性。</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><img src="https://mmbiz.qpic.cn/sz_mmbiz_png/IJTyfegB2HtQESI9ZJ4N5IqsI3ic5sEaQCYfUaiaJPCF4lXbKnBdbgRxIibgfMLgy2bp3pqb1dibZR20kO1UiaZ7qsw/640?wx_fmt=png&from=appmsg&wxfrom=5&wx_lazy=1&wx_co=1&&tp=webp" style="width: 50%; margin-bottom: 20px;"></span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><strong style="color: blue;">PART01 <span style="color: black;">科研</span>介绍</strong></span></p><span style="color: black;">在数字化时代，用户体验（UX）<span style="color: black;">评定</span><span style="color: black;">作为</span>了<span style="color: black;">保证</span><span style="color: black;">制品</span>质量和用户满意度的关键环节。然而，传统的UX<span style="color: black;">评定</span><span style="color: black;">办法</span>，尤其是对可用性测试视频的分析，<span style="color: black;">常常</span>既耗时又耗费资源。随着人工智能（AI）技术的飞速发展，<span style="color: black;">科研</span>人员<span style="color: black;">起始</span>探索<span style="color: black;">怎样</span>利用AI来辅助UX<span style="color: black;">评定</span>，以提<span style="color: black;">有效</span>率和准确性。在这一背景下，主动式对话人工智能助手（CAs）<span style="color: black;">做为</span>一种新兴的工具，展现出了在UX<span style="color: black;">评定</span>中<span style="color: black;">供给</span>实时反馈和<span style="color: black;">意见</span>的<span style="color: black;">潜能</span>。这些AI助手能够<span style="color: black;">经过</span>自然语言处理与UX<span style="color: black;">评定</span>者进行交互，从而在<span style="color: black;">评定</span>过程中<span style="color: black;">供给</span>辅助。</span><span style="color: black;">本文探讨主动式对话AI助手在UX<span style="color: black;">评定</span>中的应用，<span style="color: black;">尤其</span>是它们在<span style="color: black;">区别</span>时间点<span style="color: black;">供给</span><span style="color: black;">意见</span>对评</span><span style="color: black;">估者分析性能和主观感受的影响。<span style="color: black;">科研</span>的核心问题是：在可用性问题<span style="color: black;">出现</span>之前、同步时和之后<span style="color: black;">供给</span>自动<span style="color: black;">意见</span>，哪种<span style="color: black;">机会</span>更能有效<span style="color: black;">提高</span>UX<span style="color: black;">评定</span>的效率和质量？</span><span style="color: black;">为了回答这一问题，<span style="color: black;">科研</span>者们设计了一项混合型Wizard-of-Oz实验，<span style="color: black;">经过</span>模拟AI助手与UX<span style="color: black;">评定</span>者的互动，来深入理解AI在UX<span style="color: black;">评定</span>中的辅助<span style="color: black;">功效</span>。</span><span style="color: black;"><span style="color: black;">文案</span>提出了以下两大<span style="color: black;">科研</span>主题：</span><span style="color: black;">1.AI自动生成<span style="color: black;">意见</span>的时间<span style="color: black;">怎样</span>影响UX<span style="color: black;">评定</span>者？</span> <span style="color: black;">1.1 分析性能（例如问题数量）</span> <span style="color: black;">1.2 主观感受（例如，效率、用户信任）</span><span style="color: black;">2. 收到AI自动生成的<span style="color: black;">意见</span>后，UX<span style="color: black;">评定</span>者<span style="color: black;">怎样</span>反应？</span> <span style="color: black;">2.1 对这些<span style="color: black;">意见</span>做出<span style="color: black;">回复</span>（例如，同意或<span style="color: black;">区别</span>意）</span> <span style="color: black;">2.2 <span style="color: black;">评定</span>这些<span style="color: black;">意见</span>的质量（例如一致程度、完整性）</span><strong style="color: blue;"><span style="color: black;">PART02</span><span style="color: black;"><span style="color: black;">科研</span><span style="color: black;">办法</span>与过程</span></strong><span style="color: black;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">为了深入探索主动式对话人工智能助手（CAs）在用户体验（UX）<span style="color: black;">评定</span>中的<span style="color: black;">功效</span>，本<span style="color: black;">科研</span>采用了混合型Wizard-of-Oz实验<span style="color: black;">办法</span>。这种<span style="color: black;">办法</span>结合了AI技术和真人<span style="color: black;">干涉</span>，以模拟AI助手与UX<span style="color: black;">评定</span>者的互动。<span style="color: black;">科研</span>的<span style="color: black;">第1</span>步是招募了24名<span style="color: black;">拥有</span><span style="color: black;">区别</span>UX<span style="color: black;">评定</span>经验的参与者，<span style="color: black;">她们</span>将对三个<span style="color: black;">区别</span><span style="color: black;">制品</span>（一个博物馆网站、一个食品配送应用和一个虚拟现实游戏）的可用性测试视频进行分析。</p>
</span><span style="color: black;">在实验中，<span style="color: black;">科研</span><span style="color: black;">运用</span>了ChatGPT，这是一种基于大型语言模型（LLM）的生成性AI工具，来自动生成关于潜在可用性问题的<span style="color: black;">意见</span>。为了<span style="color: black;">保证</span>这些<span style="color: black;">意见</span>的质量，<span style="color: black;">科研</span><span style="color: black;">首要</span>利用Zoom的自动转录功能将可用性测试视频中的口头内容转录成文本。<span style="color: black;">而后</span>，研究人员对这些转录文本进行校正，添加标点，并编辑时间戳以<span style="color: black;">暗示</span>自然语言的停顿。接着，ChatGPT<span style="color: black;">按照</span>这些转录文本生成可用性问题的<span style="color: black;">意见</span>。</span>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/sz_mmbiz_jpg/IJTyfegB2HtvB7WHezrtFLv6Mfib7KgdoQLbDOvgO9MuXNN8hHv6mR4UFiccYR1RIb9bM8l990Y4cYESvp3ZYeFw/640?wx_fmt=jpeg&from=appmsg&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">图1 <span style="color: black;">ChatGPT 响应的屏幕截图，其中<span style="color: black;">包括</span>四个可用性问题描述以及<span style="color: black;">每一个</span>问题的<span style="color: black;">起始</span>和结束时间</span></span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">为了模拟AI助手的实时反馈，<span style="color: black;">科研</span>设计了三种<span style="color: black;">区别</span>的自动<span style="color: black;">意见</span><span style="color: black;">机会</span><span style="color: black;">要求</span>：在潜在问题之前、同步时和之后。这些<span style="color: black;">意见</span><span style="color: black;">经过</span>一个用户界面展示给参与者，该界面<span style="color: black;">包含</span>一个视频播放器和一个聊天窗口。视频播放器<span style="color: black;">准许</span>参与者审查可用性测试视频，而聊天窗口则用于展示AI助手的自动<span style="color: black;">意见</span>和对参与者问题的响应。</span></p><img src="https://mmbiz.qpic.cn/sz_mmbiz_jpg/IJTyfegB2HtvB7WHezrtFLv6Mfib7Kgdok3fOrQONBJxhfTvHibTljqWkMsVdPZsfsiapAIVwza0F47ErtbXHC7sQ/640?wx_fmt=jpeg&from=appmsg&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"><span style="color: black;">图2 </span><span style="color: black;"><span style="color: black;">视频时间轴说明了三种时序<span style="color: black;">要求</span>：1) <span style="color: black;">意见</span>出<span style="color: black;">此刻</span>问题之前，2) <span style="color: black;">意见</span>与问题同步<span style="color: black;">显现</span>，3) <span style="color: black;">意见</span>出<span style="color: black;">此刻</span>问题之后</span></span><span style="color: black;">在分析过程中，参与者<span style="color: black;">能够</span><span style="color: black;">经过</span>聊天窗口对AI助手的自动<span style="color: black;">意见</span>做出反应，<span style="color: black;">包含</span><span style="color: black;">暗示</span>同意、<span style="color: black;">需求</span>澄清、提出纠正或<span style="color: black;">暗示</span><span style="color: black;">区别</span>意。<span style="color: black;">另外</span>，参与者还<span style="color: black;">能够</span>提出与视频内容<span style="color: black;">关联</span>的问题，由真人演员（即wizard）<span style="color: black;">装扮</span>的AI助手会即时<span style="color: black;">回复</span>这些即兴问题。这种设计<span style="color: black;">准许</span><span style="color: black;">科研</span>收集关于自动<span style="color: black;">意见</span><span style="color: black;">机会</span>对参与者分析<span style="color: black;">行径</span>影响的数据，以及参与者对AI助手<span style="color: black;">意见</span>的接受程度和质量<span style="color: black;">评定</span>。</span>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/sz_mmbiz_jpg/IJTyfegB2HtvB7WHezrtFLv6Mfib7Kgdoll8FsbE3Kiaykbic0CzZwVRLqymDzeIQykDVXrzic7QicMrFc9QxPcfd9Q/640?wx_fmt=jpeg&from=appmsg&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p><span style="color: black;">图3 </span><span style="color: black;"><span style="color: black;">UX分析工具的用户界面：(A) 视频播放器、(a1) 进度栏、(B) 聊天线程、(b1) 聊天框、(b2) <span style="color: black;">表示</span><span style="color: black;">意见</span>、(b3) <span style="color: black;">表示</span><span style="color: black;">信息</span>和 (b3)<span style="color: black;">表示</span><span style="color: black;">所有</span></span></span><span style="color: black;">最后，<span style="color: black;">科研</span>过程中收集的数据<span style="color: black;">包含</span>参与者对自动<span style="color: black;">意见</span>的反应、<span style="color: black;">她们</span>自己识别的可用性问题、对AI助手<span style="color: black;">意见</span>的同意程度以及<span style="color: black;">她们</span>对AI能力的主观感知。<span style="color: black;">经过</span>定量和定性的分析<span style="color: black;">办法</span>，<span style="color: black;">科研</span><span style="color: black;">评定</span>了<span style="color: black;">区别</span><span style="color: black;">机会</span>的自动<span style="color: black;">意见</span>对UX<span style="color: black;">评定</span>者分析性能和主观感知的影响，以及AI助手在UX<span style="color: black;">评定</span>中的潜在价值和局限性。</span><img src="https://mmbiz.qpic.cn/sz_mmbiz_jpg/IJTyfegB2HtvB7WHezrtFLv6Mfib7KgdoxTP6M7KicWicMPb8Pics0FLTbNAicGzCTAiamEV3CLevsK1UXnyCL3Uk3vA/640?wx_fmt=jpeg&from=appmsg&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"><span style="color: black;">图4 <span style="color: black;">表示</span><span style="color: black;">科研</span>程序的流程图</span><strong style="color: blue;"><span style="color: black;">PART03 </span><span style="color: black;">实验结果分析</span></strong><span style="color: black;">
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">实验结果分析部分<span style="color: black;">首要</span>关注了自动<span style="color: black;">意见</span><span style="color: black;">机会</span>对UX<span style="color: black;">评定</span>者分析性能的影响。<span style="color: black;">经过</span>对比三种<span style="color: black;">区别</span><span style="color: black;">机会</span><span style="color: black;">要求</span>——在潜在问题之前、同步时和之后——对参与者识别问题数量的统计，<span style="color: black;">科研</span><span style="color: black;">发掘</span><span style="color: black;">机会</span><span style="color: black;">要求</span>并未<span style="color: black;">明显</span>影响参与者<span style="color: black;">发掘</span>问题的总数。这<span style="color: black;">显示</span>，尽管自动<span style="color: black;">意见</span><span style="color: black;">能够</span><span style="color: black;">供给</span><span style="color: black;">帮忙</span>，但<span style="color: black;">评定</span>者的核心分析能力并未因AI的介入而受到<span style="color: black;">明显</span>改变。</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">其<span style="color: black;">科研</span>深入探讨了参与者对自动<span style="color: black;">意见</span>的主观感知。<span style="color: black;">经过</span>调查和访谈，<span style="color: black;">科研</span>者<span style="color: black;">发掘</span>大<span style="color: black;">都数</span>参与者更偏好在潜在问题<span style="color: black;">出现</span>后接收自动<span style="color: black;">意见</span>。这种偏好的<span style="color: black;">背面</span><span style="color: black;">原由</span>是，<span style="color: black;">评定</span>者倾向于<span style="color: black;">首要</span>独立分析问题，<span style="color: black;">而后</span>再利用AI的<span style="color: black;">意见</span><span style="color: black;">做为</span>验证。这种<span style="color: black;">次序</span>不仅<span style="color: black;">加强</span>了<span style="color: black;">评定</span>者对AI<span style="color: black;">意见</span>的信任，<span style="color: black;">亦</span><span style="color: black;">加强</span>了<span style="color: black;">她们</span>的工作效率。</p>
</span>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/sz_mmbiz_jpg/IJTyfegB2HtvB7WHezrtFLv6Mfib7Kgdo4S71bMyMwTrexKqHA0viaIUN3H1wPfuMvfibWVOWzgGXZibsHDIS7oYfA/640?wx_fmt=jpeg&from=appmsg&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><span style="color: black;">图5 参与者对每种<span style="color: black;">要求</span>的效率、信任和偏好的评分</span></span></p><span style="color: black;">进一步分析参与者对ChatGPT生成的自动<span style="color: black;">意见</span>的<span style="color: black;">详细</span>反应，<span style="color: black;">科研</span>揭示了参与者对建议的四种<span style="color: black;">重点</span>反应：同意、纠正、澄清请求和<span style="color: black;">区别</span>意。其中，大部分<span style="color: black;">意见</span>（77.6%）得到了参与者的同意，这一结果<span style="color: black;">显示</span>AI助手在辅助识别可用性问题方面<span style="color: black;">拥有</span><span style="color: black;">必定</span>的效用。然而，<span style="color: black;">亦</span>有相当比例的<span style="color: black;">意见</span>未被接受，这<span style="color: black;">一般</span>是<span style="color: black;">由于</span><span style="color: black;">意见</span>与<span style="color: black;">评定</span>者的观察结果不符或<span style="color: black;">意见</span>描述<span style="color: black;">不足</span>准确。</span>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/sz_mmbiz_jpg/IJTyfegB2HtvB7WHezrtFLv6Mfib7KgdovPJprnPZRB2KZ4K9s87lD7HNbvqOZeZuXxA4OLJ8MiaiakavdsRGibN6w/640?wx_fmt=jpeg&from=appmsg&tp=webp&wxfrom=5&wx_lazy=1&wx_co=1" style="width: 50%; margin-bottom: 20px;"></p><span style="color: black;">图6 </span><span style="color: black;"><span style="color: black;">参与者对每种<span style="color: black;">状况</span>的认知<span style="color: black;">奋斗</span>、满意度和<span style="color: black;">帮忙</span>性的评分</span></span><span style="color: black;"><span style="color: black;">科研</span>分析了ChatGPT在识别可用性问题上的准确性。<span style="color: black;">经过</span>与UX专家的手动分析结果对比，ChatGPT的精确度（precision）为86%，召回率（recall）为71.1%。这一结果<span style="color: black;">寓意</span>着虽然ChatGPT能够识别出<span style="color: black;">有些</span>正确的问题，但<span style="color: black;">亦</span><span style="color: black;">错失</span>了近一半的独特问题。这一<span style="color: black;">发掘</span>强调了人类专业知识在UX<span style="color: black;">评定</span>中的不可替代性，<span style="color: black;">同期</span><span style="color: black;">亦</span>指出了当前AI工具在理解和分析<span style="color: black;">繁杂</span>用户交互方面的局限性。</span><strong style="color: blue;"><span style="color: black;">PART04 </span><span style="color: black;">结论与讨论</span></strong><span style="color: black;"><span style="color: black;">科研</span>结果揭示了自动<span style="color: black;">意见</span><span style="color: black;">机会</span>对UX<span style="color: black;">评定</span>者分析性能和主观感知的<span style="color: black;">明显</span>影响。参与者对ChatGPT生成的自动<span style="color: black;">意见</span>的反应多种多样。总体上，77.6%的<span style="color: black;">意见</span>被接受，这<span style="color: black;">显示</span>AI助手在<span style="color: black;">必定</span>程度上能够辅助UX<span style="color: black;">评定</span>者识别问题。然而，<span style="color: black;">亦</span>有部分<span style="color: black;">意见</span>未被接受，这可能是<span style="color: black;">因为</span><span style="color: black;">意见</span>的描述<span style="color: black;">不足</span>准确<span style="color: black;">或</span>与参与者的分析结果不一致。<span style="color: black;">另外</span>，<span style="color: black;">科研</span>还<span style="color: black;">发掘</span>，尽管AI助手能够识别出<span style="color: black;">有些</span>可用性问题，但它<span style="color: black;">错失</span>了58.8%的由参与者集体识别的独特问题。这一结果<span style="color: black;">明显</span>了人类专业知识在UX<span style="color: black;">评定</span>中的不可替代性，尤其是在理解用户<span style="color: black;">行径</span><span style="color: black;">背面</span>的<span style="color: black;">繁杂</span>性和上下文方面。AI助手<span style="color: black;">日前</span>还<span style="color: black;">没法</span>完全替代人类<span style="color: black;">评定</span>者的洞察力和经验，<span style="color: black;">因此呢</span>，<span style="color: black;">将来</span>的AI工具应该被视为人类<span style="color: black;">评定</span>者的辅助工具，而不是替代品。</span><span style="color: black;">在讨论部分，<span style="color: black;">科研</span>者们提出了对<span style="color: black;">将来</span>AI辅助UX<span style="color: black;">评定</span>工具的设计和功能的<span style="color: black;">意见</span>。例如，<span style="color: black;">能够</span><span style="color: black;">思虑</span><span style="color: black;">准许</span>UX<span style="color: black;">评定</span>者<span style="color: black;">按照</span>个人<span style="color: black;">爱好</span><span style="color: black;">调节</span>自动<span style="color: black;">意见</span>的<span style="color: black;">机会</span>，以优化<span style="color: black;">她们</span>的分析流程。<span style="color: black;">另外</span>，<span style="color: black;">将来</span>的AI工具<span style="color: black;">能够</span>利用多模态数据（如视频内容、用户的表情和语音）来<span style="color: black;">加强</span>对可用性问题的识别能力。<span style="color: black;">科研</span>者们还强调了在AI工具中<span style="color: black;">知道</span>其局限性和<span style="color: black;">运用</span>指南的重要性，以<span style="color: black;">保证</span><span style="color: black;">评定</span>者能够有效地利用AI助手，<span style="color: black;">同期</span>保持对<span style="color: black;">评定</span>结果的独立判断。这项<span style="color: black;">科研</span>为理解AI助手在UX<span style="color: black;">评定</span>中的潜在<span style="color: black;">功效</span><span style="color: black;">供给</span>了宝贵的见解，并为<span style="color: black;">将来</span>人机协作工具的发展<span style="color: black;">供给</span>了<span style="color: black;">指点</span>。<span style="color: black;">经过</span>优化AI助手的设计，<span style="color: black;">能够</span>使其<span style="color: black;">作为</span>UX<span style="color: black;">评定</span>者的辅助工具，从而<span style="color: black;">加强</span><span style="color: black;">评定</span>的效率和质量。</span>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;">注</p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">本文是CHI中的一篇论文《<strong style="color: blue;"><span style="color: black;">Enhancing UX Evaluation Through Collaboration with Conversational AI Assistants: Effects of Proactive Dialogue and Timing</span></strong>》。</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">原文链接：<span style="color: black;">https://dl.acm.org/doi/10.1145/3613904.3642168</span>编者<span style="color: black;">根据</span>读书笔记的形式进行撰写，原论文知识产权归杂志社<span style="color: black;">或</span>原论文作者所有，如有侵权，请联系删除。</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"> 封面<span style="color: black;">照片</span><span style="color: black;">源自</span>：<span style="color: black;">https://i.pinimg.com/originals/dd/31/83/dd3183eebb116baf5e252a92a540d316.jpg</span>（如有侵权，请联系删除）</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"> i读 </span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">“i读”<span style="color: black;">经过</span>紧跟时事热点、分享国内外设计<span style="color: black;">行业</span><span style="color: black;">优秀</span>论文、论文撰写经验与iDi项目介绍，让读者在平时的点滴<span style="color: black;">累积</span>中深入理解、<span style="color: black;">把握</span>设计<span style="color: black;">科研</span><span style="color: black;">办法</span>，<span style="color: black;">加强</span>学术<span style="color: black;">科研</span>能力与论文写作水平</span></p>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;">◆ </span><span style="color: black;">◆ </span><span style="color: black;">◆ <span style="color: black;">◆ </span><span style="color: black;">◆</span></span></p><span style="color: black;">编辑：皓淳 俊瑶 </span><span style="color: black;">编辑责任人：凡睿</span><img src="https://mmbiz.qpic.cn/sz_mmbiz_jpg/IJTyfegB2HuN7FnAibstANXF2YqsyjpR9GpSyHpqHicSiaT2Zgws1oWOLDPSDuibRSug8coZwnIqHjkj0WCzFOnhmQ/640?wx_fmt=jpeg&wxfrom=5&wx_lazy=1&wx_co=1&&tp=webp" style="width: 50%; margin-bottom: 20px;"><strong style="color: blue;"><span style="color: black;">— <span style="color: black;">举荐</span>阅读 —</span></strong><span style="color: black;">（点击蓝字<span style="color: black;">就可</span>阅读）</span><a style="color: black;"><span style="color: black;">湖南大学智能设计与交互体验实验室简介</span></a><a style="color: black;"><span style="color: black;">赵江洪 工业设计学科“教授的教授”</span></a>
<h1 style="color: black; text-align: left; margin-bottom: 10px;"><a style="color: black;"><span style="color: black;">iDi实验室两篇论文入选ESI 2023高被引论文</span></a></h1><a style="color: black;">iDi实验室邹一鸣、付熙、叶丹澜、欧阳汭霖论文获UXPA2022文集最佳论文奖</a><a style="color: black;">习惯性</a>和成瘾性智能手机<span style="color: black;">行径</span>与<span style="color: black;">那些</span><span style="color: black;">原因</span><span style="color: black;">关联</span>呢？
<h1 style="color: black; text-align: left; margin-bottom: 10px;"><a style="color: black;">车内语音交互，越“快”越好？</a></h1>
<p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><img src="https://mmbiz.qpic.cn/sz_mmbiz_png/IJTyfegB2Hu9rQIUeibfSdnWM1lkeSD5Pic1nYnDAzwwNxDtYiaUBHcibpEkQFdG9h4D9OM1dseEXibeGqV4icFMnsibA/640?wx_fmt=png&from=appmsg&wxfrom=5&wx_lazy=1&wx_co=1&&tp=webp" style="width: 50%; margin-bottom: 20px;"></p><span style="color: black;"><span style="color: black;">倘若</span>有收获，记得戳</span><span style="color: black;"><strong style="color: blue;"><span style="color: black;">在</span></strong><strong style="color: blue;"><span style="color: black;">看</span></strong></span><span style="color: black;">哦</span>
“沙发”（SF，第一个回帖的人） 楼主继续加油啊！外链论坛加油！ 你字句如珍珠，我珍藏这份情。
页:
[1]