文章主题：
这次介绍一本小书：The art of asking ChatGPT。这本书介绍了使用不同的提示技巧（Prompt techniques）从ChatGPT获取高质量回答的方法。作者使用简单易懂的语言，结合实际的解释和示例，对每种提示技巧进行了详细的讲解。书中还提供了将不同提示技巧组合以实现更具体结果的示例。而且，很多时候具体的例子比抽象的公式很有力，所以本文就每个技巧直接给出例子，相信不用做过多解释，都会比葫芦画瓢。不管你是普通人、研究人员、开发人员还是想将ChatGPT作为个人助手的人，这本书都适合。希望你会喜欢并从中受益。
ChatGPT使用英文互动的准确性远比用中文强很多，所以很多例子均以英文进行，并主要通过给出具体例子来学习提问技巧。有关中文网络内容的数量和质量问题对ChatGPT或者其他AI的影响，推荐看马逆最新节目：【【睡前消息574】ChatGPT是做题家 中国欠他一套模拟卷】 https://www.bilibili.com/video/BV1Tm4y1m7ty/?share_source=copy_web
在使用ChatGPT时，了解它的能力和限制非常重要。如果没有正确的引导，模型可能无法产生所需的结果。这就是提示工程的作用，通过提供明确而具体的指令来引导模型的输出，并确保生成的文本相关性。提示公式是指提示的特定格式，通常由三个主要元素组成：Task 任务（清晰而简洁地陈述提示要求模型生成的内容）、Instruction 指令（在生成文本时模型应遵循的指令）和 Role 角色（在生成文本时模型应扮演的角色），有时还可以挤上关键字（Seed-word）。本书总共提供了23个技巧，非常有实操价值。
技巧1: 指令型提示（Instructions Prompt Technique）
清晰简洁的任务 + 具体的指令
公式：”Generate [task] following these instructions: [instructions]”
例子：“Generate professional and accurate responses to customer inquiries following these instructions: The responses should be professional and provide accurate information.“
实际操作中表现很不错：
技巧2: 角色型提示（Role Prompting）
公式： “Generate [task] as a [role]”
例子：“Generate responses to customer inquiries as a customer service representative.”
也可以结合技巧1进行提示：
“As a marketing representative, generate an informative, persuasive product description that highlights the innovative features of the new smartphone. The smartphone has the following features [insert your features]”
技巧3：标准提示（Standard prompts）
公式：”Generate a [task]“
例子：”Generate a summary of this news article“
Standard prompts, role prompting, and seed-word prompting techniques提示技巧相结合的例子：“As a tech expert, generate an objective and informative product review that highlights the powerful features of the new laptop.“
技巧4：零范例、一个范例和少数范例的提示
公式：”Generate text based on [number] examples”
✨Introducing the Future Smartwatch – Your Ultimate Timeless Companion✨💻Experience the epitome of sophistication and technology with our sleek, all-new smartwatch. Crafted to perfection, it’s not just a watch; it’s an extension of your digital lifestyle. 📱Featuring cutting-edge features like advanced health tracking, seamless connectivity to your smartphone, and personalized smart notifications, this watch keeps you connected without any hassle.🎨Aesthetically pleasing with a choice of premium finishes, the Future Smartwatch boasts a minimalist design that blends seamlessly into any outfit. Its customizable straps ensure a perfect fit for every wrist.⚡Stay ahead of the game with its long battery life, ensuring you never have to worry about charging during your busy day. 🌍Whether you’re traveling or just staying active, this watch is equipped with GPS and multi-sport tracking, making it your go-to companion for all adventures.🎁Don’t miss out on the opportunity to upgrade your style and enhance your daily routine. Get ready to revolutionize your timekeeping game with the Future Smartwatch – the ultimate smartwatch for the modern era.SEO optimized: “Revolutionize Your Timepiece with the Future Smartwatch – A Sleek, Advanced Watch for Every Lifestyle”
🌟📱最新智能手机对比🔥：iPhone 13 vs. 独家新品🔥👀 当科技巨头们再次亮剑，这款全新智能手机无疑吸引了所有目光。今天，我们就来一场尖端对决，看看它如何与苹果的热销旗舰iPhone 13一较高下！💥🔍 首先，设计上两者可谓各有千秋。iPhone 13延续了经典的玻璃边框和流畅过渡，而我们的新品则大胆创新，采用一体化金属材质，更显现代感与高级质感。🎨🔋 动力方面，iPhone 13搭载A15仿生芯片，性能强劲，但续航依旧；而我们这款产品，电池容量升级，待机时间更长，让你告别频繁充电的困扰。⚡️📷 摄像头是决定手机摄影体验的关键。iPhone 13虽有光学防抖和强大的Leica合作，但在像素上并未突破；而我们的新品，采用独家研发的高解析度镜头，夜景模式下效果惊人，让拍照更具艺术感。📸💰 购买决策往往伴随着价格考量。虽然两者在性能上有相当竞争，但我们的产品定价更为亲民，性价比超乎想象。💰总结来说，这款新品凭借独特的设计和出色的续航，以及超越iPhone 13的摄影体验，无疑为消费者提供了更多选择。是时候考虑升级你的科技生活了！🚀#智能手机对比# #iPhone13对比# #新品亮点
🌟📖最新电子阅读器评测📚🌟如果你正在寻找一款集时尚与功能于一体的电子阅读设备，那么这款新推出的e-reader绝对不容错过！它以其卓越的阅读体验和创新设计，在众多竞品中脱颖而出。让我们一起看看它如何在同行中独树一帜吧！首先，这款e-reader的屏幕清晰度超乎想象，无论是细腻的文字还是鲜艳的图片，都能呈现出令人惊艳的效果。它的护眼模式更是贴心至极，长时间阅读也不会感到眼睛疲劳。相比之下，那些色彩泛滥、字体模糊的设备就显得黯然失色了。设计方面，它轻巧便携，大小适中，握持感舒适，无论是放在包包里还是床头柜旁，都能轻松融入你的日常生活。而相比那些厚重且操作复杂的设备，这款e-reader的操作流畅，界面简洁，简直是阅读党的福音！当然，我们也不能忽视它的续航能力。一次充电，能让你畅读数周，无需频繁插拔，节省了大量时间。相比之下，有些型号的电子书阅读器电池寿命短得让人头疼。最后，我们选取了几款市场上的热门e-reader进行对比：Kindle Oasis、Kobo Forma和Nook GlowLight 3。虽然它们各有千秋，但这款新e-reader在价格与性能上都展现出相当的优势，性价比极高。如果你还在犹豫不决，不妨亲自体验一下，相信它会成为你的阅读首选！总结来说，这款新e-reader凭借其出色的显示效果、人性化设计和超长续航，无疑已经站稳了电子书阅读器的前沿。想要提升阅读品质，就不要错过这个绝佳的机会！📚🎉
技巧5：”Let’s think about this”提示
这种技术适用于写作、诗歌或创意写作等任务，生成具有反思性和沉思性的文本。
Task: Write a reflective essay on the topic of personal growth
公式例子:
“Lets think about this: personal growth”；
“Lets think about the impact of climate change on agriculture”；
“Lets discuss the current state of artificial intelligence”；
“Lets talk about the benefits and drawbacks of remote work”
技巧6：一致性型提示 （Self-Consistency Prompt）
🌟ChatGPT一致性保障🌟ChatGPT的强大在于其对文本内容的一致性把控，无论是严谨的事实核查，还是细致的数据验证，它都能游刃有余地执行。这项技术在文本生成领域中扮演着关键角色，确保了信息的准确性和连贯性。🚀无论你需要检查文档中的事实准确性，还是确认数据集的整体一致性，ChatGPT都是你信赖的得力助手。🛡️无需担心信息错漏，让ChatGPT帮你扫清障碍，确保每个细节都精准无误。
例子：”Generate a product review that is consistent with the following product information [insert product information]”
“Summarize the following news article in a way that is consistent with the information provided [insert news article]”
“Complete the following sentence in a way that is consistent with the context provided [insert sentence]”
“📊Check this stat – July’s avg temp hits a high of 30°C, but the lowest reading clocks in at a comfortable 20°C. 🌊Consistency maintained across the board! 📊”
技巧7：关键字提示（Seed-word prompt）
通过提供特定的关键词或短语来控制ChatGPT的输出，往往和其他技巧结合。
“Please generate text based on the following seed-word: Dragon”
“Generate a poem related to the seed word “love” and should be written in the style of a sonnet.”
“Complete the following sentence in a way that is related to the seed word science and in the style of a research paper as a researcher: [insert sentence]”
技巧8：知识生成型提示（Knowledge Generation prompt）
从ChatGPT获取新的和原始的信息。
当然可以！🎉 当涉及到特定主题时，比如你对XYZ的兴趣🔥，我可以为你创作一篇独特且有价值的内容。这将包括深入探讨该话题的各个方面，并确保信息准确无误🔍。我会用精准的语言和恰当的结构来吸引搜索引擎的眼球SEO优化，同时保持内容的连贯性和易读性📖。让我们一起探索这个主题，让你获取最新、最相关的信息！🔥
“Generate new and accurate information about [specific topic] ““Answer the following question: [insert question]”“Integrate the following information with the existing knowledge about [specific topic]: [insert new information]”“Please generate new and original information about customer behavior from this dataset”
技巧9：知识整合型提示（Knowledge Integration prompts）
整合新信息或连接不同的信息。这种技术有助于将现有知识与新信息相结合，以产生对特定主题的更全面的理解。
“Integrate the following information with the existing knowledge about [specific topic]: [insert new information]”
“Connect the following pieces of information in a way that is relevant and logical: [insert information 1] [insert information 2]”
“Update the existing knowledge about [specific topic] with the following information: [insert new information]”
技巧10：多项选择型提示（Multiple Choice prompts）
有特定的问题，或有要特定的内容作为潜在答案。此技术可用于生成仅限于一组特定选项的文本，并可用于问答、补充文本和其他任务。例子：
“Answer the following question by selecting one of the following options: [insert question] [insert option 1] [insert option 2] [insert option 3]””Complete the following sentence by selecting one of the following options: [insert sentence] [insert option 1] [insert option 2] [insert option 3]””Classify the following text as positive, neutral or negative by selecting one of the following options: [insert text] [positive] [neutral] [negative]”
技巧11：可解释的软提示（Interpretable Soft Prompts）
允许一定的模糊空间和灵活性来控制生成的文本。需要提供特定输入和一些所需输出的来完成的。
“Generate a story based on the following characters: [insert characters] and the theme: [insert theme]““Complete the following sentence in the style of [specific author]: [insert sentence]“
“Generate text in the style of [specific period]: [insert context]“
技巧12：生成受控型提示（Controlled Generation prompts）
对输出文本具有高度的可控制性。这是通过为模型提供一组特定的输入来实现的，例如模板、特定的词汇表或一组条件，用于指导生成的过程。
“Generate a story based on the following template: [insert template]““Complete the following sentence using the following vocabulary: [insert vocabulary]: [insert sentence]“
“Generate text that follows the following grammatical rules: [insert rules]: [insert context]“
技巧13：问答型提示（Question-answering prompts）
生成回答特定问题或任务的文本。通过向模型提供具体问题或任务，以及可能与问题或任务相关的任何其他信息来实现的。
“Answer the following factual question: [insert question]”
“Define the following word: [insert word]”“Retrieve information about [specific topic] from the following source: [insert source]”
技巧14：总结型提示（Summarization prompts）
对文本或者书籍文章做总结
“Summarize the following news article in one short sentence: [insert article]”“Summarize the following meeting transcript by listing the main decisions and actions taken: [insert transcript]”“Summarize the following book in one short paragraph: [insert book title]”
技巧15：对话型提示（Dialogue prompts）
通过为模型提供上下文和一组特征，以及它们的角色和背景，并要求模型在它们之间生成对话。
“Generate a conversation between the following characters [insert characters] in the following context [insert context]”
“Generate a dialogue between the following characters [insert characters] in the following story [insert story]”
“Generate a professional and accurate dialogue for a customer service chatbot, when the customer asks about [insert topic]”
技巧16：对抗型提示（Adversarial prompts）
这个技巧有些难翻译，感觉总是无法对应到一个比较自然的中文词语，暂时翻译成对抗性提示吧。总之，这个技巧是为了，生成某些比较难以生成的文本类型。具体意思请揣摩一下三个例子：
“Generate text that is difficult to classify as [insert label]”“Generate text that is difficult to classify as having the sentiment of [insert sentiment]”“Generate text that is difficult to translate to [insert target language]”
技巧17：聚类型提示（Clustering prompts）
聚类型提示是根据某些特征将类似的数据点组合在一起的技术。这种技术对数据分析、机器学习和自然语言处理等任务很有用。
“Group the following customer reviews into clusters based on sentiment: [insert reviews]”“Group the following news articles into clusters based on topic: [insert articles]”“Group the following scientific papers into clusters based on research area: [insert papers]”
技巧18：强化学习提示（Reinforcement learning prompts）
允许模型从其过去的行动中学习，并随着时间的推移提高其性能。提示还应包括有关所需输出的信息，例如要完成的任务以及任何具体要求。这种技术对决策、游戏和自然语言生成等任务很有用。
“Use reinforcement learning to generate text that is consistent with the following style [insert style]”“Use reinforcement learning to translate the following text [insert text] from [insert language] to [insert language]”“Use reinforcement learning to generate an answer to the following question [insert question]”
技巧19：课程学习型提示（Curriculum learning prompts）
通过逐渐增加难度来学习复杂的任务。提示还应包括所需输出的信息，例如要完成的最终任务以及任何具体要求。这种技术适用于自然语言处理、图像识别和机器学习等任务。使用方法和上面的技巧类似。
“Use curriculum learning to generate text that is consistent with the following styles [insert styles] in the following order [insert order]”“Use curriculum learning to translate text from the following languages [insert languages] in the following order [insert order]”
“Use curriculum learning to generate answers to the following questions [insert questions] in the following order [insert order]”
技巧20：情绪分析型提示（Sentiment analysis prompts）
确定一段文本的情感基调或态度，例如它是积极的、消极的还是中立的。应向模型提供一段文本，并要求其根据其情绪对其进行分类。提示还应包括有关所需输出的信息，例如要检测到的情绪类型（例如积极、消极、中立）以及任何特定要求或约束。这种技术对自然语言处理、客户服务和市场研究等任务很有用。
“Perform sentiment analysis on the following customer reviews [insert reviews] and classify them as positive, negative, or neutral.”“Perform sentiment analysis on the following tweets [insert tweets] and classify them as positive, negative, or neutral.”“Perform sentiment analysis on the following product reviews [insert reviews] and classify them as positive, negative, or neutral.”
技巧21：命名实体识别提示（Named entity recognition prompts）
在文本中识别和分类已命名的实体，如人员、组织、地点和日期。提示还应包括有关所需输出的信息，例如要识别的命名实体类型（例如人员、组织、地点、日期）以及任何特定要求或约束。
“Perform named entity recognition on the following news article [insert article] and identify and classify people, organizations, locations, and dates.”“Perform named entity recognition on the following legal document [insert document] and identify and classify people, organizations, locations, and dates.”“Perform named entity recognition on the following research paper [insert paper] and identify and classify people, organizations, locations, and dates.”
技巧22：文本分类型提示（Text classification prompts）
将文本分类为不同的类别或类别。这种技术对自然语言处理、文本分析和情感分析等任务很有用。需要注意的是，文本分类与情感分析不同。情绪分析特别侧重于确定文本中表达的观点或情感。这可以包括确定文本是否表达了积极、消极或中立的情绪。情绪分析通常用于客户评论、社交媒体帖子和其他表达情感很重要的文本。
“Perform text classification on the following customer reviews [insert reviews] and classify them into different categories such as electronics, clothing and furniture based on their content.”
“Perform text classification on the following news articles [insert articles] and classify them into different categories such as sports, politics, and entertainment based on their content.”
“Perform text classification on the following emails [insert emails] and classify them into different categories such as spam, important, or urgent based on their content and sender.”
技巧23：文本生成型提示（Text generation prompts）
文本生成提示与本书中提到的其他几种提示技术有关，例如：Zero, One and Few Shot Prompting, Controlled generation prompts, Translation prompts, Language modeling prompts, Sentence completion prompts. 所有这些提示都涉及生成文本，但它们在生成文本的方式和对生成文本的特定要求或约束方面有所不同。文本生成提示可用于微调预训练模型或为特定任务训练新模型。
“Generate a story of at least 1000 words, including characters [insert characters] and a plot [insert plot] based on the following prompt [insert prompt].”“Translate the following text [insert text] into [insert target language] and make sure that it is accurate and idiomatic.”“Complete the following text [insert text] and make sure that it is coherent and consistent with the input text.”Chapter 26: Word prediction prompts结语合适的Prompt是获得高质量答案的必要条件。通过输入包含各种技巧的提示，我们可以指导模型生成适合特定需求的文本。我们要以不同的方式使用这些技巧中的一种或多种结合，找到最适合的那个。
AI时代，掌握AI大模型第一手资讯！AI时代不落人后！
免费ChatGPT问答，办公、写作、生活好得力助手！
扫码右边公众号，驾驭AI生产力！
相关文章
