2025-06-20 | Google Make AI Work for You | A Practical Guide to Using AI for Work
用AI提升工作效率:掌握提示技巧与实用工具
标签
媒体详情
- 上传日期
- 2025-06-20 18:23
- 来源
- https://www.youtube.com/watch?v=a-fmRIW8u-8
- 处理状态
- 已完成
- 转录状态
- 已完成
- Latest LLM Model
- gemini-2.5-pro
转录
speaker 1: What do you love about your work? Maybe it's coming up with creative ideas, connecting with your customers or getting a project over the finish line with your team. But there might be the parts that you don't love so much, like managing your inbox or your ever growing to do list. And often those things can get in the way. Fortunately, AI can help you handle those time consuming tasks, but when and how do you use it for your job or business? That's exactly what I'm here to show you right now. I want you to ask yourself the following questions. What's a task you wish you could delegate this week? What's a task you've been putting off because you just don't know where to begin? And what's a repetitive or tedious task that makes your day to day life less enjoyable? Pause the video if you need to, and make note of your answers. As you go through this course, you're going to explore how AI can help with those challenges, and you'll leave with a plan to get started. My name is Wendy, and I've been working with small businesses at Google for over a decade. We are at such an exciting time as we all start to integrate AI power tools into our work. And my goal is to help make sure that everyone is able to benefit. That includes small businesses and workers across jobs and industries. There is so much opportunity, but as with anything new, knowing where to start can be challenging. In this course, I'll guide you as you practice using AI in different ways to tackle some of your daily challenges. We'll focus on Google Gemini and notebook lm, but what you learn here can be used with other AI experiences too. But here's what this course is not about. I'm not going to tell you to add to your to do list. It's all about using AI to help you with the things you're already doing. We'll dive into different examples that you can apply right away, and I'll demonstrate how you can get the best results when using AI. I'm also going to encourage you to use AI instead of just learning about it. Only knowing AI in theory would be like reading about how to play the piano without ever touching the keys. So I'll often ask you to pause the video and practice with me. You can also check out the activities in the video description that will help you practice what you learned. Here's what you can expect in this course. You'll get to know how AI works in everyday terms, not all that tech jargon you may have heard before. Next, you'll discover how to write effective prompts, which is basically how to give AI the best instructions to get the most useful results. You'll be prompting along with me the whole time, so get ready to experiment as you learn. Once you're warmed up, we'll cover how to enhance your creativity with Gemini and Gemini live, how to get help from an AI assistant right in the middle of tasks you're already doing with Google workspace, how you can better understand your information with notebook lm. At every step, you'll discover real examples of how people are using AI in their day to day lives. We'll show you how you can apply their ways of working with AI to your own tasks and challenges. Now, before you move on, follow the link in the video description to help you identify some of the most time consuming tasks you're dealing with right now. Then stick around to learn how AI can help accomplish them more quickly and easily. Like me, you've probably been using AI for years already without knowing it. When I commute to work, real time traffic updates let me know what routes to avoid. If I'm struggling to write an email, AI suggests the perfect phrase. AI is also what powers the recommendations I get for videos on YouTube, and those are only a few examples. But how can AI predict what welike to watch, write, or do? I'm going to quickly break this down for you, because knowing how AI works can help you understand when to use it. Let's start with an AI experience. You might be familiar with, Gemini. It's available as an AI assistant on your desktop or phone. You can also access Google workspace with Gemini in apps like gmail, docs, slides, or sheets. Gemini is Google's most capable AI model yet, and it's based on years of experience developing AI technology and building it into your favorite Google products like search, maps, and pixel. Right now you might be thinking, wait, what's an AI model? Simply put, it's a computer program that takes a bunch of data, learns from it, and uses those learnings to make predictions, recommendations or decisions. In a way, Gemini and other AI models are kind of like a chef creating a recipe. Their first attempt is one data point. From that data, they learn things like, the recipe needs more salt. Their following attempts add more data points and learnings, like it's too spicy, it needs vegetables and so forth. In the end, all these learnings help the chef decide on the perfect recipe. This type of learning ability is what makes AI so helpful. Now imagine our chef has been learning from thousands of recipes they've created, and they've seen how people have responded to different food combinations. The chef now has enough knowledge to create a brand new food that they predict people will love, like combining a donut and croissant into an innovative new pastry like the chef. Once an AI model has learned from massive amounts of data, it can start doing pretty impressive things. It can take a request and come up with a response, even if it has never encountered such a request before. This type of AI model is called a large language model, or llm for short. Lllens can learn so much and be so good at predicting that they can create entirely new content like blog posts, budgets, or project plans. This is called generative AI or gen AI for short. Some gen AI assistants, like the Gemini app, don't only work with text. Gen AI can understand and create images, videos, audio and more. So you could upload a photo of a chart and ask for a tecsummary. You could get a list of action items from an audio recording of a meeting, or you could generate striking visuals for your website or client presentation. Okay. So how do we actually use Gemini? Well, it comes down to how we communicate with it. We do this through prompts, basically the instructions we give AI to get the results we want. Think of it like telling a really helpful assistant what you want them to do. You might have heard AI technology can sometimes hallucinate. That's a strange term, but it basically just means that AI sometimes gets things wrong or says things that aren't. So it's your responsibility to always use a human in the loop approach when working with AI, meaning you need to verify AI results before you use them in your work. And there's your super quick overview of how AI works. Now that you understand the basics, you're ready to apply AI to your own work tasks and challenges. Imagine you ask someone directions to a restaurant. They tell you, make a left, another left, and then a right. You probably won't feel super confident that you'll get where you need to go, right? But if that person had given you more precise instructions, like make a left on main street at the corner with the gas station and continue for two blocks until you reach the restaurant, you're more likely to find where you're going to get useful results from AI. It's similar. You need to give it clear and specific guidance to head in the right direction. Communicating with AI is called prompting, and it's different from the way we usually engage with technology. You're not just giving commands. You're engaging in a conversation, one that you can build on and refine to get exactly what you need. In the best part, these conversations happen in natural language, which means you don't need special skills to get started. You're ready right now. So it all comes down to being a good communicator. One of the biggest mistakes people can make is not including enough details in their prompts. You need to be clear and specific when you prompt Gemini or any gen AI I experience to get the best results. Generally speaking, the clearer and more specific you are with the prompts you put in, the better results you're going to get out. People usually call the results from gen AI outputs, and our goal is to get you the best ones that help you solve your challenges. Now I'm going to show you how to write effective prompts. It's a good time to hit pause and open up Gemini so that you can prompt alongside me. The first step to effective prompting is describing the task you want Gemini to help you with. Take a moment to think of a task you wish you could delegate right now. If you're not quite sure which task is a fit, here's a little tip. Pick something that involves creating something new, like a marketing plan or blog post. Or it could be changing something you've already created, like rewriting a product description for a different audience. If you still can't think of a task to tackle, try a version of the prompt I'm about to show you. Let's imagine I've just opened a floral design business and need some customers. I want to reach out to local businesses to Spark their interest in partnering with me. So I open Gemini and describe the task I want to delegate. Help me write an introductory email to local businesses about my new floral design business with the goal of exploring potential partnerships. You do the same. Tell Gemini what you wanted to help you with, but don't hit enter yet. We're going to keep building out this prompt so that Gemini can get as close as possible to what we're looking for. The second step is adding a persona to the prompt. This tells Gemini what expertise or knowledge to draw from when approaching your task. For example, if you need help revising a timeline for a project, you could tell Gemini to act as a project manager who never misses a deadline. Or if you need copywriting support, you could tell Gemini to take on the persona of a creative writer at a successful publication. Act as an email marketer who specializes in new business development. Add a persona to your own prompt now, but don't hit enter. There are still three more things you may want to include. Let's move on to the third step, which is specifying the format. This means telling Gemini what the output should look like. Do you want na receive the output as bullet points or a list of pros and cons? Do you want a specific length or tone? Do you want it in another language? Type in a format for your prompt and I'll do the same. The output should be 500 words max, with a friendly tone, bullet points for readability, and an engaging subject line to drive open rates. The fourth step is setting the context, which is any background information Gemini needs to know to help you with your task. The more context you give, the better your output will be. You should know that I source flowers locally, specialize in custom arrangements and offer bespoke pricing for loyal customers. Add some context to your prompt. Now think about sharing your goals, the reason for the task, or even what you've tried before but didn't work. Just one more step to go. So resist the temptation and don't hit enter yet. The fifth and final step is adding references. This tells Gemini what to aim for and why. Imagine you have an intern who is writing a Press release for your company for the first time. This would be like giving them examples of Press releases you like or don't like to guide them in the right direction. References work similarly for Gemini. Let's say I just got a promotional email from a local bakery that really grabbed my attention. I want Gemini to use it as a reference, so I'll attach it to my prompt and include instructions on how to use it. Use the attached email as a model for its persuasive style. You might not have any references available for your task right now, but if you do, feel free to add them to your prompt. If you've kept up with me and written your own prompt, you can finally hit enter. But we're not done yet. Part of being a responsible AI user is evaluating your outputs before using them. I'd never send an email to potential clients without reading it first. I'll review my output and I'll ask myself, does it meet my goals? Is it accurate and useful? Let's see. Jim and I did its job. I asked for a friendly email that was no more than 500 words long. I also wanted a Gemini to highlight a few things, locally sourced flowers and bespoke pricing. So far, so good. Now evaluate your output. Did you get good results? If so, great. In many cases, making the most of AI usually involves refining or building on your initial prompt, a process we call iteration. For example, I might iterate on my email prompt and ask Gemini for a shorter email with fewer bullet points. Okay. So you just built an effective prompt for Gemini. You included a task, persona and format. You added some context and maybe a reference. Keep in mind that while we follow these steps in a specific order, the substance is what really matters. And here's another tip. If you write a prompt that gives you a great output, save it. The next time you're doing a similar task, you can reuse your prompt instead of starting from scratch. One thing I like to do is save my favorite prompts in a Google doc. Then all I have to do is paste them into Gemini and update the details for my new prompt. This is a great way to build a prompt library or a collection of go to prompts that make working with Gemini faster and easier. To help you get started on yours, check out the prompt library linked in the video description. If you like the prompt you've just written, copy it there and keep adding the prompts that work best for you. Now you know how to prompt effectively. You can apply what you've learned to any AI assistant or model. If you need additional help delegating your tasks to Gemini, follow the link in the video description. If you found this video helpful, hit subscribe below. For more AI learning, think of a famous building and everything that went into its design. The Eiffel Tower didn't go from idea to fully finished monument in just one try. It probably started with sketches, experimentation and refinement. In the end, these design tweaks resulted in an iconic tower that's strong enough to last centuries. This is how you should approach AI prompting. It's not a one and done effort. If your initial AI outputs aren't quite right, you can learn from them and try different approaches to your prompt until you get the right results. Anytime you rework your initial prompt or follow it up with more guidance. For Gemini, that's iteration, and there are many ways to do it. One basic iteration method is making sure your original prompt includes a task, persona, format, context, and references. Add in whatever's missing or add more detail to what you already have. Another way to iterate is breaking down a large prompt into several smaller ones, taking Gemini through the process step by step. Instead of asking for a full marketing plan, first ask Gemini to help you develop a way to position your product and then develop a plan to acquire customers. You can also iterate by switching to a similar but slightly different task. For example, let's say you prompted Gemini to build a marketing strategy for your lawn care business, but you're not getting a good result. You can iterate by asking for a step by step plan to grow your customer base for your lawn services. One final iteration I recommend is adding constraints or limitations, like specifying a workout or asking that your plan focus on what can be done in the next month for less than $500. Now let's practice improving a prompt through iteration here. Gemini has just helped me write an email introducing my floral design business to potential partners, but it feels a little too wordy for someone to quickly scan, so I'll iterate by asking Gemini to make its output more concise. Thank you for your help. This is a great start, but it feels too long for an introductory email, please shorten to 250 words or less. Now, you're probably familiar with text based prompts, but what if you could go beyond text, like include different types of media in your prompt? That's right. Gemini understands more than just words. You can attach pdf's, documents, pictures, code, and more to your prompts. Think about how you communicate with your hairdresser. You could describe the haircut you want using words, but wouldn't a photo be way more effective? This technique is called multimodal prompting, and I'll show how it works by continuing to iterate on my prompt. Imagine I want to customize my email for a specific recipient. So I'm going to add a new piece of context to my prompt, the flyer of a business I want to partner with. All I have to do is attach the image to my prompt, analyze the attached flyer, then personalize my introductory email for this specific business. And just like that, Gemini helps me customize my email. Think about all the ways you could use multimodal prompting. You could turn a photo of a storage closet into an inventory list, turn a recording of a lecture into a study guide, connect your calendar and ask for help with time management and so much more. We've just covered a lot of ground on how to iterate on your prompts. Now give it a try yourself. If you're looking for additional help on iteration, follow the link in the video description. We all hit points in our workday where we get stuck, where a quick brainstorm or a second opinion can make all the difference. Whether tackling a tricky decision or just needing fresh inspiration, being able to bounce ideas around or get extra input can be invaluable. But let's face it, that perfect sounding board isn't always available. This is where AI can step in. You can use Gemini as a creative thought partner, ready to assist with all kinds of tasks. It can help you come up with solutions, work more efficiently and quickly create and transform content instead of having a one sided exchange with AI, you're engaging in a dialogue that will help you think in new ways and enhance your creativity. So what's the best way to partner with Gemini to think things through? The answer is prompt chaining. This technique guides Gemini through a series of interconnected prompts, adding new layers of context or tasks every step of the way. Each new prompt expands on a specific part of the previous output. This can be really effective for identifying solutions to a problem, comparing different options, or turning ideas into a plan. I'm going to open up Gemini and show you how prompt chaining works. To follow along, take a moment to think about a challenge where you could use a thought partner. Is there somewhere you need help getting off the starting block, or are you trying to find creative ways to tackle a problem? Take a second to pause the video. Open Gemini and follow along. If you can't think of a task of tackle, try a version of the prompt I'm about to show you in my scenario, imagine I own a painting company and when to expand by offering additional services. I need some ideas, so I enter a simple prompt to start the brainstorming process. I want to expand my painting comp's service offerings, act as my business advisor, and list three complimentary services we could provide. Gemini gives me three suggestions and explains why each service is complementary to painting. I'm interested in these options, but I really need something I can offer as soon as possible. Time is of the essence, so I'll ask Gemini to recommend my fastest option. Which of these can we launch within six months? Okay, power washing it is. Notice how Gemini tells me why the other two options might take a little longer, which really helps me make my decision. So I'm going with power washing, but how can I get from idea to implementation? Gemini can help with that too. If we want to move forward, what tasks should I complete this week? And there we go. Gemini gives me a detailed breakdown of each action item and the rationale for recommending it. So with just three prompts, I've gone from a pretty broad idea around wanting to expand my business to a concrete plan for making it happen. That's the power of prompt chaining. You can start with something abstract and ask Gemini to help you turn it into something tangible. And here's a pro tip. Gemini saves your conversations for later. So whenever you wanna revisit the same task or keep brainstorming, just open up the same chat and pick up right where you left off. Now you might be wondering, how is prompt chaining different from iteration? With iteration, you refine your request over and over to get better help with the same task. It's like asking an associate at a hardware store, how do I fix a leaky faucet? The associate gives you instructions. You then iterate by saying, okay, thanks. But can you explain that again step by step? Like I've never done this before? You're getting the answer to the same question, but refined to be more helpful to you. With prompt chaining, you use the output from one prompt as a building block for your next one. Instead of focusing on one task and refining, you're building a series of interconnected tasks. I'll explain using our hardware store analogy. First, you ask the sales associate, how do I fix a leaky faucet? Then you might ask, okay, now that I know the steps, what specific tools will I need and which brands do you recommend? Once you've gathered your tools, you may follow up with what safety precautions should I take? Remember, with iteration you're improving the output, whereas with prompt chaining, you're taking the output and building upon it for a new related request. There are so many ways Gemini can help you in your day to day work. Now be sure to check out other videos in this series to see how real people are doing it and how you can too. If you found this video helpful, hit subscribe below and share it. Have you ever sat down to start typing out a long email and then thought, hang on a second, wouldn't it be so much faster and easier to get on a call and have an actual conversation? Sometimes it's easier to just talk it out. In these moments, you can go live with Gemini. Gemini live is a conversational experience where you can talk with Gemini to get real time responses. No typing required. To start, download the Gemini app on your phone and click on the Gemini live button. You're still using Gemini as your creative thought partner, but you're working through it live. There are so many other ways to use Gemini live. It's great for brainstorming ideas or getting quick advice, but you can also use it to simulate conversations like rehearsing a sales pitch, a presentation, or tough conversation. Gemini live can assess your delivery and give you tips to improve it as you go, which has some real advantages over practicing in front of the mirror. Gemini live is multimodal, too. In addition to using your voice as an input, you can use your phone's camera to have Gemini identify objects, solve problems, and understand what you're experiencing in real time, so it can help you troubleshoot your broken printer, take a quick stock ckroom inventory, or refine your product display. You can even share your phone screen to get instant feedback on whatever you're working on, like your latest social media post or your website's new layout. Gemini live analyzes your content, identifies areas for improvement, and gives you personalized recommendations on the spot. Let me show you how it works. Imagine I run a mailing and shipping company. I want to tell repeat customers about a new promotion we're offering. But first I decided to go live and practice my pitch. To launch Gemini live. I'll open the Gemini app on my phone, tap the live icon and start speaking Gemini. Let's role play a scenario. speaker 2: Sounds fun. What kind of scenario are you thinking of? speaker 1: You're a repeat customer ster of my shipping company, and I'm calling to let you know about some special promotions for running. Pretend to pick up the phone and just respond naturally like you're a busy but polite business owner. speaker 2: Sure thing. Just a moment. Let me get into character. Okay, I'm ready. speaker 1: Hi there. It's Wendy from the shipping company you used last year. How are you doing? speaker 2: Oh, hi, Wendy. I remember you. I'm doing well. Thanks for asking. It's been a busy year for us, but things are good. What can I do for you today? speaker 1: I'm sure you're super busy, but I wanted to let you know about some discounts we're offering right now just in case you have any upcoming shipping needs. speaker 2: That's very thoughtful of you, Wendy. I appreciate you reaching out. We do have some shipments coming up, so I'm definitely interested. What kind of . speaker 1: discounts are we talking about? So up until the end of next month, you can get 20 off all shipping jobs, and we're offering 10% off on prepaid mailing boxes. Is this something yoube interested in . speaker 2: that sounds promising? We do have several large shipments coming up, and we're always looking for ways to save on costs. Could you tell me a little more about the prepaid mailing boxes? What sizes do you offer and what are the price . speaker 1: points with the discount? Gemini, let's pause the role play. Wanna give me some feedback on my delivery? speaker 2: Okay, sure thing. Overall, I thought you did a good job with staying in character and keeping the conversation flowing. Your voice had a friendly and professional tone, which is perfect for a customer service call. There were a couple of spots where the script felt a little too salesy. Maybe try to sound a bit more conversational and less like you're reading from a script. speaker 1: Okay, so not only did Gemini help me rehearse, but I got some great suggestions on what to improve when you're generating new text or image content or uploading documents to your prompt, you'll probably want to stick with the Gemini desktop experience. But if you prefer to have a free flowing conversation or talk through a problem on the go, try using Gemini live. There are so many ways Gemini live can help you on the go. Be sure to check out other videos in this series to see how real people are solving real problems at work and how you can too. How many apps or tabs do you have open on your computer phone right now? If you're anything like me, it can be so many that I can't even see what I have open. It's probably because you have a lot of information tools and web apps living in different places, and you're constantly moving from one to the next. What if you could simplify this digital juggling act? This is where Google workspace with Gemini comes in. It helps you make the most of Gemini right inside the apps you're already using each day. And if you tell it to, Gemini can tap into your documents and files to understand exactly what you're working on. Think about a recent project that took a lot of time and energy. Maybe you worked on a staff memo and Google Docs, or a client presentation in slides, or a quarterly sales report in sheets. Now imagine having an AI assistant in that document presentation or spreadsheet with you, making you even more productive and efficient. Google workspace already centralizes your essential business tools, making it easy to collaborate with others and share your work. But with Gemini, you can ask for AI's help without breaking your flow or opening another tab. Remember, whenever you're writing a prompt for AI, the more context you give, the better can help you. Prompting Gemini within docs, slides, or sheets gives Gemini the context it needs. Instead of having to copy and paste information or upload documents into the Gemini app, you can prompt Gemini right inside whatever workspace app you're working in. Take it from me. I'm not great at making slides, and if I need to include images, it takes me an embarrassing amount of time to find the right one. But with Gemini and slides, it's so much easier. I just describe what I'm looking for, and Gemini creates that image directly on the slide for me. No more time wasted searching for the perfect image. I create exactly what I need without opening another tab within Google workspace. Gemini can help with other tasks like summarizing long email threads in gmail, creating proposals for a new project in docs, analyzing data in sheets, generating talking points for a meeting and slides, taking notes and providing meeting summaries in meet, and quickly finding and synthesizing all of this information in your Google Drive. Now's a good time to pause this video and give one of these a try. Open up a dock or a sheet you're working in, find the Gemini Spark in the top right corner, and experiment. There are so many ways Gemini can assist you within the workspace apps you're already working in. Stick around for more videos in this series to see how real people are using it and how you can too. I love receiving an email that opens with tldr or too long didn't read this term has become work slang for a quick summary of a long document, project, or idea. Receiving a tldr is great. I'm able to stay informed without much time or effort. But what about creating a tldr? And what if you need to pull that summary from all kinds of sources, documents, articles, websites, even videos? This is where notebook lm comes in. Feed at your notes, your documents, or even your favorite articles and itinstantly give you insightful summaries, connect related ideas, and even help you generate new content based on what you've given it. It's like having a supercharged tldr machine. But for everything you need to understand, I am the type of person who writes everything down. I use poits emails, calendar holds in countless open tabs. The intention is to go back to these information nuggets, but since they're all over the place, it's really hard to do. But now with notebook lm, I'm not only able to organize all of this information in one place, it actually helps me do something with all of the notes I'm saving. See, what's unique about notebook lm is that it only focuses on the stuff you give it, like documents, videos, websites, and presentations. You create a new notebook and upload or link your own sources. If you want, you can also ask it to pull in relevant information from across the web using the discover sources feature. Then you can ask questions about the information in your notebook. Notebook lm can help recap key points, make connections among topics, and uncover patterns you might have missed. This saves you time on research and makes you super efficient at analyzing documents and turning raw data into insights. Give it a try. Open up notebook lm and upload some sources. Maybe you're interested in learning more about trends in your industry or understanding what gives your competitor an edge. Or maybe you want to quickly get up to speed on a project without having to read through a bunch of different documents. Once resources are uploaded, ask a few questions to get a feel for how it works, or choose one of the suggested questions notebook lm automatically gives you. Okay, so you may be thinking, what challenges can notebook lm help me solve? And how is that different from Gemini or Google workspace with Gemini? Let's break it down. Gemini generates new content based on all the data it's been trained on, which can enhance your creativity and act as a thought partner. Similarly, Google workspace with Gemini generates new content based on all of its training data, but this AI assistant helps you right in the middle of tasks you're already doing. Since you're working within a workspace app, Gemini has plenty of context to understand what you need, and you don't need to open a new tab to use it. Notebook m generates new content based on only the data you've fit. Its outputs are laser focused on your sources, helping you understand your information better. Think of it like this. Gemini is like going to your local library, where you can borrow any of the books on the shelves. But notebook lm is like your home library. You're selecting things to learn about from your own personal bookshelf. While it may have fewer books than the public library, they're all ones you've personally acquired. And one of my favorite things that has been a game changer. Notebook lm can turn all your written content into an audio overview. Audio overviews are lively, in depth discussions between two AI hosts who summarize your materials, make connections between topics, and banter back and forth. Many of us learn and remember better by listening to conversations, so this can be a great way for you and your team to digest complex information without needing to look at a screen. You can even download an audio overview and listen to it in the same way youlisten to a podcast during your commute, at the gym or while you're folding laundry. There are so many ways notebook lm can help you better understand your information and act on it. Be sure to check out more videos in this series to see how real people are using it to solve their everyday challenges and how you can too. Congratulations on finishing make AI work for you. Now it's time to start making AI a habit. As you've learned, AI is something you can integrate into your workflow and benefit from right away. Just bring your challenges and let AI help you solve them. You don't need a background in tech to use AI successfully, nor do you have to completely overhaul the way you work to get the most out of it. So the next time you sit down to do something and think, this is the last thing I want to do right now, or I really don't have time for this, or I feel stuck, I want you to ask yourself, is this something AI can help with? Make it a habit to examine what you're already doing and find ways to make AI work for you. And if you're ever unsure which AI solution is right for your needs or how it can help you with a challenge, just ask Gemini for guidance. Here's my personal challenge to you. Identify three ways you're going to use AI in the next day. Give it a try. Experiment, play. The best way to start getting the benefits of this new technology is by doing to help you with this ongoing journey, use the worksheet found in the video description to build your action plan. It's designed to help you think about a specific challenge youlike to tackle with AI, and it will help you plan your approach, choose the right AI solution, and evaluate your results. This is your personal roadmap for integrating AI into your daily work. If you found this video helpful, hit subscribe below for more AI tips.
最新摘要 (详细摘要)
概览 (Executive Summary)
本课程由Google的Wendy主讲,是一份面向各类工作者与小企业的实用指南,旨在阐释如何将AI工具(特别是Google Gemini和NotebookLM)应用于日常工作,以解决耗时、重复或棘手的任务。课程的核心理念是,AI并非遥不可及的技术,而是通过有效的沟通技巧——即“提示工程 (Prompting)”——便能即刻利用的强大助手。通过掌握编写高效提示词的方法,任何人都能将AI转变为提升效率、激发创意的“思想伙伴”。
课程系统性地讲解了编写高效提示词的五步框架:任务、人设、格式、背景和参考。同时,课程介绍了迭代 (Iteration) 和提示词链 (Prompt Chaining) 等高级技巧,用以持续优化AI的输出质量。此外,课程对Google生态中的关键AI工具及其应用场景进行了区分和说明:
* Gemini:作为通用的创意引擎和问题解决伙伴。
* Gemini Live:提供基于语音和摄像头的实时、动态对话式交互。
* Google Workspace中的Gemini:将AI无缝嵌入Docs、Sheets等日常办公应用,在工作流中提供即时支持。
* NotebookLM:作为用户专属资料的深度分析工具,能对指定的文档、视频等内容进行精准的总结、关联和洞察生成,并提供创新的“音频概览”功能。
最终,课程倡导将使用AI培养成一种工作习惯,鼓励用户通过持续的实践探索,将AI无缝融入现有工作流程,从而更智能、高效地完成任务。
第一部分:AI基础与工作原理
- AI的本质与类比:AI已融入日常生活(如地图交通预测、邮件智能撰写)。其核心是AI模型——一种通过学习海量数据来进行预测或决策的计算机程序。课程用一个生动的比喻来解释这个过程:AI模型就像一位厨师,最初的尝试是一个数据点,通过不断学习反馈(“需要更多盐”、“太辣了”)来积累经验,最终创造出完美的食谱。
- 大型语言模型 (LLM) 与生成式AI (Gen AI):当AI模型学习了足够多的数据后,便能像那位经验丰富的厨师创造全新菜品一样,生成前所未见的全新内容(如博客文章、预算方案)。这种模型被称为大型语言模型 (LLM),其创造新内容的能力则被称为生成式AI (Gen AI)。
- 多模态能力:先进的生成式AI(如Gemini)不仅能处理文本,还能理解和创建图像、视频、音频等多种格式的内容。例如,用户可以上传图表照片要求AI生成文本摘要,或从会议录音中提取行动要点。
- 核心风险与用户责任:
- AI幻觉 (Hallucination):AI有时会提供错误或不实的信息。
- “人在环路”原则 (Human in the Loop):这是使用AI时一项至关重要的责任。用户必须对AI生成的所有结果进行人工核实与审查,确保其准确性后再投入使用。
第二部分:核心技能:编写高效AI提示词
向AI下达清晰、具体的指令是获得高质量输出的关键。课程提出了一个包含五个关键要素的提示词框架,并鼓励用户将有效的提示词保存起来,建立个人“提示词库”以便复用。
- 1. 任务 (Task):清晰描述希望AI完成的具体工作。
-
示例:“帮我写一封介绍我新花艺业务的邮件,目标是与本地商家建立合作。”
-
- 2. 人设 (Persona):为AI指定一个专家角色,使其从特定视角提供帮助。
-
示例:“扮演一个擅长新业务开发的邮件营销专家。”
-
- 3. 格式 (Format):规定输出内容的具体形式,如长度、语气、结构(要点列表)或语言。
-
示例:“输出应在500字以内,语气友好,使用要点以提高可读性,并包含一个吸引人的主题行。”
-
- 4. 背景 (Context):提供所有必要的背景信息,帮助AI更好地理解任务。
-
示例:“你应该知道,我的花卉是本地采购的,我擅长定制花艺,并为老客户提供专享价。”
-
- 5. 参考 (References):提供正面或反面的示例,为AI的输出提供模仿或规避的基准。
-
示例:“使用附件中的邮件作为其说服力风格的范本。”
-
第三部分:高级技巧与应用策略
- 迭代 (Iteration):对初始提示词进行修改或追加指导,以逐步优化输出。这是一个反复调整直至满意的过程。
- 主要方法:补充缺失的提示词要素;将复杂任务拆解为多个小任务;微调任务方向;增加预算、时间等约束条件。
- 多模态提示 (Multimodal Prompting):在提示词中结合文本以外的媒介(如PDF、图片、代码),以提供更丰富的上下文。
-
示例:上传一张商家传单的图片,并要求Gemini基于该传单内容,为这家企业个性化一封推广邮件。
-
- 提示词链 (Prompt Chaining):指创建一系列相互关联的提示词,后一个提示词建立在前一个输出的基础上,从而引导AI逐步完成一个多步骤的复杂任务。
- 与迭代的区别:迭代是为同一任务优化输出(“请把这个回答解释得更简单”);而提示词链是用一个输出作为新相关任务的输入(“既然知道了步骤,我需要哪些工具?”)。
第四部分:Google AI工具应用场景解析
4.1 Gemini:全能创意伙伴
将与AI的交互从单向问答升级为双向对话,利用其进行头脑风暴、获取第二意见。通过提示词链技术,可以引导Gemini将一个抽象的想法(如“我想拓展业务”)逐步转化为具体的行动计划。
4.2 Gemini Live:实时语音交互
Gemini移动应用中的一项功能,允许用户通过语音与AI进行实时、自然的对话。
* 主要用途:
* 对话模拟:排练销售演讲或重要谈话,并获得关于语气、措辞的即时反馈。
* 实时视觉辅助:结合手机摄像头,让AI识别物体、诊断设备故障或优化产品陈列。
* 即时内容反馈:通过分享手机屏幕,让AI对社交媒体帖子、网站布局等提供改进建议。
4.3 Google Workspace中的Gemini:无缝集成的工作流助手
将AI助手直接嵌入用户日常使用的Google应用(如Docs, Sheets, Slides)中,无需切换标签页即可获得帮助。Gemini能自动将当前文档内容作为上下文,提供高度相关的支持。
* 应用示例:在Gmail中总结长邮件、在Docs中起草项目提案、在Slides中根据文字描述直接生成图片、在Meet中自动生成会议纪要。
4.4 NotebookLM:专属资料的深度分析师
这是一款专注于处理和分析用户所提供资料的AI工具。其知识范围被严格限定在用户上传的来源(文档、网站、视频等)之内,确保输出结果的精准性和相关性。
* 核心类比:Gemini如同公共图书馆,知识广博;NotebookLM则是你的私人书房,知识精准且聚焦于你的个人收藏。
* 特色功能:音频概览 (Audio Overview):将书面资料转换成一段生动的双人AI主播对话。他们会以讨论、辩论的形式总结你的材料,非常适合听觉型学习者或希望在通勤等场景下消化复杂信息的用户。
结论:将AI融入日常,化为工作习惯
课程的最终目标是鼓励用户将AI视为一种日常工具,并养成在遇到挑战时主动思考“AI能否帮忙?”的习惯。课程强调“学以致用”,通过发起“一天内用AI解决三个问题”的挑战,并提供行动计划工作表,号召用户立即开始实践,将AI的潜力真正转化为个人的生产力。