nqkk58 发表于 2024-9-28 11:10:27

怎么样运用人工智能来办事:AI实用指南


    <h1 style="color: black; text-align: left; margin-bottom: 10px;">随着人工智能技术的快速发展,<span style="color: black;">各样</span>强大的AI系统正在以日新月异的速度<span style="color: black;">面世</span>。上周,<span style="color: black;">咱们</span>见证了Claude 2的首次亮相,它可能是<span style="color: black;">日前</span>公众能接触到的第二强大的AI系统。再上<span style="color: black;">1星期</span>,OpenAI发布了Code Interpreter,<span style="color: black;">日前</span>为止最<span style="color: black;">繁杂</span>的AI模式。再上<span style="color: black;">1星期</span>,<span style="color: black;">有些</span>AI<span style="color: black;">得到</span>了“看<span style="color: black;">照片</span>”的能力。</h1>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><span style="color: black;">然而,似乎<span style="color: black;">无</span>一个AI实验室<span style="color: black;">供给</span>任何用户文档。<span style="color: black;">日前</span>能看到的用户指南,都只是来自Twitter意见领袖的讨论线程。针对声<span style="color: black;">叫作</span>关心自己技术适当<span style="color: black;">运用</span>的组织<span style="color: black;">来讲</span>,<span style="color: black;">选取</span>以传言形式进行文档化是一种奇特的<span style="color: black;">选取</span>,但<span style="color: black;">咱们</span>还是处在这种状态。</span></span></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;">我<span style="color: black;">不可</span>声<span style="color: black;">叫作</span>这会是一个完整的用户指南,<span style="color: black;">然则</span>它将为当前的AI<span style="color: black;">情况</span><span style="color: black;">供给</span><span style="color: black;">有些</span>方向感。我<span style="color: black;">始终</span>在为学生(和感兴趣的读者)准备AI入门指南,每隔几个月就需要作出重大修改。<span style="color: black;">近期</span>的几个月格外疯狂。</span></span></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;">本指南表达个人观点,基于我的经验,专注于</span><span style="color: black;"><span style="color: black;">怎样</span><span style="color: black;">选取</span>正确的工具来完成任务</span><span style="color: black;">。<span style="color: black;">倘若</span>你想先<span style="color: black;">认识</span>AI<span style="color: black;">能够</span>完成<span style="color: black;">那些</span>任务,我之前写的<span style="color: black;">文案</span>可能会有<span style="color: black;">帮忙</span>。</span></span></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;"><span style="color: black;">重点</span>的大型语言模型</span></span></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;">当<span style="color: black;">咱们</span><span style="color: black;">此刻</span>讨论AI时,<span style="color: black;">一般</span>是在讨论大型语言模型(LLM)。大<span style="color: black;">都数</span>AI应用都<span style="color: black;">是由于</span>LLM<span style="color: black;">供给</span>支持的,<span style="color: black;">日前</span><span style="color: black;">仅有</span><span style="color: black;">少许</span>几家组织创建了<span style="color: black;">基本</span>模型。每家<span style="color: black;">机构</span>都<span style="color: black;">经过</span>聊天<span style="color: black;">设备</span>人<span style="color: black;">供给</span>对其模型的直接<span style="color: black;">拜访</span>:OpenAI制造了GPT-3.5和GPT-4,它们支持</span><span style="color: black;">ChatGPT</span><span style="color: black;">和Microsoft的</span><span style="color: black;">Bing</span><span style="color: black;">(在Edge浏览器上<span style="color: black;">能够</span><span style="color: black;">拜访</span>它);谷歌<span style="color: black;">持有</span><span style="color: black;">各样</span>以</span><span style="color: black;">Bard</span><span style="color: black;">为标签的模型;Anthropic推出了Claude和</span><span style="color: black;">Claude 2</span><span style="color: black;">。</span></span></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;"><span style="color: black;">这儿</span>不会讨论其他LLM。<span style="color: black;">第1</span>个<span style="color: black;">是由于</span>Inflection创建的Pi,它针对对话进行了优化,真的非常非常想<span style="color: black;">作为</span>你的<span style="color: black;">伴侣</span>(说真的,试试就<span style="color: black;">晓得</span>我的意思了)。它除了聊天之外不想做太多事,想让它为你工作将会非常<span style="color: black;">熬煎</span>。<span style="color: black;">咱们</span><span style="color: black;">亦</span>不会涵盖<span style="color: black;">各样</span>开源模型,任何人都<span style="color: black;">能够</span><span style="color: black;">运用</span>和修改。<span style="color: black;">日前</span>它们<span style="color: black;">针对</span>普通用户<span style="color: black;">来讲</span><span style="color: black;">一般</span>不可<span style="color: black;">拜访</span>且毫无用处,<span style="color: black;">然则</span><span style="color: black;">将来</span>指南可能会<span style="color: black;">包含</span>它们,<span style="color: black;">由于</span>它们确实<span style="color: black;">特别有</span>前景。</span></span></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;"><span style="color: black;">因此</span>这是你的快速参考表,总结了LLM的当前状态:</span></span></p>
    <div style="color: black; text-align: left; margin-bottom: 10px;"><img src="https://p3-sign.toutiaoimg.com/tos-cn-i-qvj2lq49k0/4b0eaa8fe2ff43c2b32ff0ac37f5b871~noop.image?_iz=58558&amp;from=article.pc_detail&amp;lk3s=953192f4&amp;x-expires=1727527649&amp;x-signature=H20G%2FfPPxLK9hfVSCKVEB03aHZY%3D" style="width: 50%; margin-bottom: 20px;"></div>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><span style="color: black;">前四个(<span style="color: black;">包含</span>Bing)都是OpenAI系统。<span style="color: black;">日前</span>OpenAI<span style="color: black;">重点</span>有两种AI:3.5和4。3.5模型在去年11月<span style="color: black;">诱发</span>了当前的AI狂潮,4模型在今年春天首次亮相,能力更强。一个新变种<span style="color: black;">运用</span>插件连接到互联网和其他应用程序。有<span style="color: black;">非常多</span>插件,大<span style="color: black;">都数</span>都不是非常有用,<span style="color: black;">然则</span><span style="color: black;">倘若</span>需要的话<span style="color: black;">能够</span>随意探索它们。代码解释器是一个极其强大的ChatGPT版本,<span style="color: black;">能够</span>运行Python程序。<span style="color: black;">倘若</span>你从未为OpenAI付费,那你只<span style="color: black;">运用</span>过3.5。除了插件版本和一个暂时中止的带有浏览功能的GPT-4版本之外,这些模型都未连接到互联网。Microsoft的Bing<span style="color: black;">运用</span>4和3.5的组合,<span style="color: black;">一般</span>是GPT-4家族中首个推出新功能的模型。例如,它<span style="color: black;">能够</span>创建和查看图像,并且<span style="color: black;">能够</span>在网页浏览器中阅读文档。它连接到了互联网。Bing的<span style="color: black;">运用</span>有点奇怪,但很强大。</span></span></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;">谷歌<span style="color: black;">始终</span>在测试自己的AI以供消费者<span style="color: black;">运用</span>,<span style="color: black;">她们</span><span style="color: black;">叫作</span>之为Bard,但它由一系列<span style="color: black;">基本</span>模型驱动,<span style="color: black;">近期</span>的一个叫做PaLM 2。<span style="color: black;">针对</span><span style="color: black;">研发</span>LLM技术的<span style="color: black;">机构</span><span style="color: black;">来讲</span>,迄今为止它们的表现相当令人失望,尽管昨天宣布的改进<span style="color: black;">表示</span><span style="color: black;">她们</span>仍在继续改进<span style="color: black;">基本</span>技术,<span style="color: black;">因此</span>我对此抱有<span style="color: black;">期盼</span>。它<span style="color: black;">已然</span><span style="color: black;">得到</span>了运行有限代码和解释图像的功能,但我<span style="color: black;">一般</span>会避免它。</span></span></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;">最后一家<span style="color: black;">机构</span>Anthropic<span style="color: black;">已然</span>发布了Claude 2。Claude最<span style="color: black;">明显</span>的特点是<span style="color: black;">持有</span>非常大的上下文窗口——基本上是LLM的记忆。Claude<span style="color: black;">能够</span>在记忆中<span style="color: black;">保留</span>近乎整本书的内容,<span style="color: black;">或</span>许多PDF。它被设计成比其他大型语言模型更不太可能产生恶意<span style="color: black;">行径</span>,在<span style="color: black;">实质</span><span style="color: black;">运用</span>中,这<span style="color: black;">寓意</span>着它倾向于对某些事情略微训斥你。</span></span></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;">接下来是<span style="color: black;">有些</span><span style="color: black;">运用</span>示例:</span></span></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;">写作</span></span></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;">最佳免费选项:</span><span style="color: black;">Bing</span><span style="color: black;">和</span><span style="color: black;">Claude 2</span></span></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;">付费选项:</span><span style="color: black;">ChatGPT 4.0</span><span style="color: black;">/带插件的ChatGPT</span></span></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;"><span style="color: black;">日前</span>,GPT-4仍然是最强大的AI写作工具,你<span style="color: black;">能够</span>在Bing(<span style="color: black;">选取</span>“创意模式”)上免费<span style="color: black;">运用</span>,<span style="color: black;">或</span>购买每月20美元的ChatGPT订阅。然而,Claude是强有力的第二<span style="color: black;">选取</span>,<span style="color: black;">况且</span>有限的免费选项可用。</span></span></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;">这些工具<span style="color: black;">亦</span>正在直接集成到常用的办公应用程序中。Microsoft Office将<span style="color: black;">包含</span>由GPT<span style="color: black;">供给</span>支持的辅导员,Google文档将集成来自Bard的<span style="color: black;">意见</span>。</span><span style="color: black;">这些创新对写作<span style="color: black;">寓意</span>着深远的影响。</span></span></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;"><span style="color: black;">这儿</span>是<span style="color: black;">有些</span><span style="color: black;">运用</span>AI<span style="color: black;">帮忙</span>你写作的方式。</span></span></p><span style="color: black;"><span style="color: black;">写任何东西的草稿。博客<span style="color: black;">文案</span>、散文、宣传材料、演讲、讲课、<span style="color: black;">选取</span>你自己的冒险、剧本、短篇小说——</span><span style="color: black;">任何你能想到的,AI都能做,<span style="color: black;">况且</span>做得很好。</span><span style="color: black;">你只需要提示它。提示创作并不是魔法,<span style="color: black;">然则</span>基本提示会产生乏味的写作,但<span style="color: black;">经过</span>互动练习提示,你会<span style="color: black;">发掘</span>AI系统<span style="color: black;">做为</span>作家的能力在稍加练习后会有很大<span style="color: black;">加强</span>。</span></span><span style="color: black;"><span style="color: black;">使你的写作更好。将文本粘贴到AI中。<span style="color: black;">需求</span>它改进内容,或<span style="color: black;">供给</span>关于<span style="color: black;">怎样</span>使其更适合特定受众的<span style="color: black;">意见</span>。<span style="color: black;">需求</span>它以10种截然<span style="color: black;">区别</span>的风格创建草稿。<span style="color: black;">需求</span>它使事物更生动,或添加示例。用它来激发你做更好的工作的灵感。</span></span><span style="color: black;"><span style="color: black;"><span style="color: black;">帮忙</span>你完成任务。AI<span style="color: black;">能够</span>做你<span style="color: black;">无</span>时间做的事情。像实习生<span style="color: black;">同样</span><span style="color: black;">运用</span>它来写电子邮件、创建<span style="color: black;">营销</span>模板、<span style="color: black;">供给</span>业务计划的后续<span style="color: black;">过程</span>等等。</span><span style="color: black;">这是我在30分钟内就能支持一次<span style="color: black;">制品</span>发布所能实现的。</span></span><span style="color: black;"><span style="color: black;"><span style="color: black;">帮忙</span>思维<span style="color: black;">起步</span>。一个<span style="color: black;">困难</span>很容易使你从任务中分心。AI<span style="color: black;">供给</span>了<span style="color: black;">得到</span>动力的方式。</span></span>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><span style="color: black;">需要担心的几件事:为了响应你的答案,AI很容易“误报”并生成<span style="color: black;">恰当</span>的虚假内容。它<span style="color: black;">能够</span>生成完全错误但<span style="color: black;">非常</span>令人信服的内容。让我强调一下:AI会<span style="color: black;">连续</span>而出色地撒谎。它告诉你的每一个事实或信息都可能不正确。你需要<span style="color: black;">检测</span>所有内容。当你请求引用、报价、引文和互联网信息时,尤其危险(<span style="color: black;">针对</span>未连接到互联网的模型)。Bing<span style="color: black;">一般</span>会比其他模型产生更少的虚构内容,<span style="color: black;">由于</span>GPT-4<span style="color: black;">广泛</span>更加基于事实,并且Bing的互联网连接<span style="color: black;">寓意</span>着它<span style="color: black;">实质</span>上<span style="color: black;">能够</span>获取<span style="color: black;">关联</span>事实。</span><span style="color: black;">这是一份避免误报的指南</span><span style="color: black;">,但不可能完全消除误报。</span></span></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;"><span style="color: black;">况且</span>请<span style="color: black;">重视</span>,AI不会解释它自己,它只会让你认为它在解释。当你<span style="color: black;">需求</span>它解释<span style="color: black;">为何</span>写了某些内容时,它会给出一个似乎<span style="color: black;">恰当</span>的完全虚构的答案。当你请求它的思考过程时,它并<span style="color: black;">无</span>审视自己的行动,它只是在生成让它看起来像是<span style="color: black;">这般</span>做的文本。这使得理解系统中的偏见非常<span style="color: black;">拥有</span>挑战性,即使这些偏见几乎肯定存在。</span></span></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;">它<span style="color: black;">亦</span><span style="color: black;">能够</span>被不道德地<span style="color: black;">操作</span>或<span style="color: black;">诈骗</span>所利用。你对其输出负有责任。</span></span></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;">生成图像</span></span></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;">最透明的选项:</span><span style="color: black;">Adobe Firefly</span></span></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;">开源选项:</span><span style="color: black;">Stable Diffusion</span></span></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;">最佳免费选项:</span><span style="color: black;">Bing</span><span style="color: black;">或</span><span style="color: black;">Bing Image Creator</span><span style="color: black;">(<span style="color: black;">运用</span>DALL-E)、Playgound(<span style="color: black;">能够</span><span style="color: black;">运用</span>多个模型)</span></span></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;">最佳图像质量:</span><span style="color: black;">Midjourney</span></span></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;">大<span style="color: black;">都数</span>人<span style="color: black;">能够</span><span style="color: black;">运用</span>的<span style="color: black;">重点</span>图像生成器有四个:</span></span></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;">Stable Diffusion,这是开源的,你<span style="color: black;">能够</span>从任何高端计算机上运行它。刚<span style="color: black;">起始</span>需要付出<span style="color: black;">奋斗</span>,<span style="color: black;">由于</span>你需要学习<span style="color: black;">怎样</span>适当地制定提示,<span style="color: black;">然则</span>一旦你<span style="color: black;">把握</span>,它就<span style="color: black;">能够</span>产生很棒的效果。它在将AI与其他<span style="color: black;">源自</span>的图像相结合方面尤其出色。</span></span></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;">OpenAI的DALL-E,它集成在Bing(必须<span style="color: black;">运用</span>创意模式)和Bing Image Creator中。该系统很不错,但不如Midjourney。</span></span></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;">Midjourney,这是2023年中期最好的系统。它的学习曲线比任何其他系统都低:只需输入“thing-you-want-to-see --v 5.2”(结尾的--v 5.2很重要,它<span style="color: black;">运用</span>的是最新模型),你就<span style="color: black;">能够</span>得到很棒的结果。</span></span></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;">Adobe Firefly,内置在<span style="color: black;">各样</span>Adobe<span style="color: black;">制品</span>中,但与DALL-E和Midjourney相比,<span style="color: black;">照片</span>质量略逊一筹。然而,<span style="color: black;">另一</span>两个模型在<span style="color: black;">运用</span>来训练AI的源图像方面模糊不清,而Adobe声明它只<span style="color: black;">运用</span>有<span style="color: black;">运用</span>权的图像。</span></span></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;">以下是它们的比较(每张<span style="color: black;">照片</span>都标注了模型):</span></span></p>
    <div style="color: black; text-align: left; margin-bottom: 10px;"><img src="https://p3-sign.toutiaoimg.com/tos-cn-i-qvj2lq49k0/4442d786828349eeb9d4ffa406c8d083~noop.image?_iz=58558&amp;from=article.pc_detail&amp;lk3s=953192f4&amp;x-expires=1727527649&amp;x-signature=y6MmqKZMLUvnirF%2BGxq%2FsVNqqvI%3D" style="width: 50%; margin-bottom: 20px;"></div>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><span style="color: black;">Prompt: “Fashion photoshoot of sneakers inspired by Van Gogh” - the first images that were created by each model</span></span></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;">提示:“梵高风格网球鞋时尚照” - <span style="color: black;">每一个</span>模型生成的<span style="color: black;">第1</span>张<span style="color: black;">照片</span></span></span></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;">需要担心的事项:这些系统<span style="color: black;">创立</span>在对互联网数据的训练上,训练本身就存在固有偏见(例如,<span style="color: black;">倘若</span>让它创造企业家的<span style="color: black;">照片</span>,默认你会看到<span style="color: black;">更加多</span>以男性为主的<span style="color: black;">照片</span>,除非你<span style="color: black;">知道</span>指定“女企业家”),你可以<span style="color: black;">运用</span>这个资源管理器来观察这些偏见。</span></span></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;">这些系统<span style="color: black;">亦</span>以在互联网上存在的艺术<span style="color: black;">做为</span>训练材料的方式进行训练,这种方式在法律和伦理上都不透明,可能存在问题。尽管技术上你<span style="color: black;">持有</span>所创造<span style="color: black;">照片</span>的版权,但法律规定仍然模糊。</span></span></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;"><span style="color: black;">况且</span><span style="color: black;">此刻</span>,它们不生成文字,只生成一堆看起来像文字的东西。<span style="color: black;">然则</span>Midjourney制作手的效果非常逼真。</span></span></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;">提出想法</span></span></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;">最佳免费选项:</span><span style="color: black;">Bing</span></span></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;">付费选项:</span><span style="color: black;">ChatGPT 4.0</span><span style="color: black;">,但<span style="color: black;">因为</span>其互联网连接,Bing可能会更好</span></span></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;">尽管存在所有约束和奇怪之处,但AI在产生想法方面完美无瑕。要有好的想法,你<span style="color: black;">一般</span>需要有<span style="color: black;">非常多</span>想法,而AI在创造<span style="color: black;">海量</span>想法方面表现不俗。<span style="color: black;">经过</span>恰当的提示,你还<span style="color: black;">能够</span>强制它进行非常创造性的思考。在Bing的创意模式下,要它<span style="color: black;">查询</span>你最<span style="color: black;">爱好</span>的不寻常的思维激发技巧,<span style="color: black;">例如</span>Brian Eno 的倾斜策略或 Mashall McLuhan 的四分法,并应用它们。<span style="color: black;">或</span><span style="color: black;">需求</span><span style="color: black;">有些</span>奇思妙想,<span style="color: black;">例如</span>从随机专利或你最<span style="color: black;">爱好</span>的超级英雄中<span style="color: black;">得到</span>灵感......</span></span></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;">制作视频</span></span></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;">最佳动画工具:</span><span style="color: black;">D-iD</span><span style="color: black;">,用于在视频中制作脸部动画。</span><span style="color: black;">Runway v2</span><span style="color: black;">,从文本创建视频</span></span></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;">最佳语音克隆:</span><span style="color: black;">ElevenLabs</span></span></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;"><span style="color: black;">此刻</span>,用完全由AI生成的角色、完全由AI编写的脚本、AI生成的语音,<span style="color: black;">经过</span>AI动画生成一个视频<span style="color: black;">已然</span>变得轻而易举。</span><span style="color: black;">它还<span style="color: black;">能够</span>深度伪造<span style="color: black;">名人</span>,正如你在这个链接中看到我深度伪造自己的那样。</span><span style="color: black;">说明和<span style="color: black;">更加多</span>信息请见<span style="color: black;">这儿</span>。<span style="color: black;">小心</span><span style="color: black;">运用</span>,但这对解释视频和介绍<span style="color: black;">特别有</span><span style="color: black;">帮忙</span>。</span></span></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;"><span style="color: black;">第1</span>个公开可用的文本到视频工具<span style="color: black;">亦</span><span style="color: black;">刚才</span>发布,即Runway v2。它创建4秒的短片段,更像是<span style="color: black;">将来</span>发展的演示,<span style="color: black;">然则</span><span style="color: black;">倘若</span>你想<span style="color: black;">认识</span>这个<span style="color: black;">行业</span>的<span style="color: black;">将来</span>发展,它值得一看。</span></span></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;">需要担心的事项:深度伪造是一个巨大的问题,这些系统需要被道德地<span style="color: black;">运用</span>。</span></span></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;">处理文档和数据</span></span></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;"><span style="color: black;">针对</span>数据(以及你有的任何奇思妙想代码):Code Interpreter(代码解释器)</span></span></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;"><span style="color: black;">针对</span>文档:Claude 2用于大型文档或多文档,Bing侧栏用于较小的文档和网页(Edge浏览器中的侧栏<span style="color: black;">能够</span>“看到”浏览器中的内容,让Bing<span style="color: black;">运用</span><span style="color: black;">哪些</span>信息,尽管上下文窗口的<span style="color: black;">体积</span>是有限的)</span></span></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;">Code Interpreter(代码解释器),它是GPT-4的一种模式,让你<span style="color: black;">能够</span>将文件上传给AI,<span style="color: black;">准许</span>AI编写和运行代码,并让你下载AI<span style="color: black;">供给</span>的结果。它<span style="color: black;">能够</span>执行程序、运行数据分析(尽管你需要足够<span style="color: black;">认识</span>统计和数据来<span style="color: black;">检测</span>它的工作)、创建<span style="color: black;">各样</span>文件、网页,<span style="color: black;">乃至</span>游戏。尽管自它发布<span style="color: black;">败兴</span>,围绕未受训人员<span style="color: black;">运用</span>它进行分析的<span style="color: black;">危害</span>存在<span style="color: black;">非常多</span>争论,但许多测试代码解释器的专家都对它印象深刻,一个论文<span style="color: black;">乃至</span><span style="color: black;">意见</span>它将需要改变<span style="color: black;">咱们</span>培训数据<span style="color: black;">专家</span>的方式。<span style="color: black;">倘若</span>你想<span style="color: black;">认识</span><span style="color: black;">更加多</span><span style="color: black;">运用</span>细节,请参考我之前的<span style="color: black;">文案</span>。我<span style="color: black;">亦</span>制作了一个初始提示,为代码解释器设置以创建有用的数据可视化。它<span style="color: black;">供给</span>了<span style="color: black;">有些</span>良好的图表设计基本原则,并提醒它<span style="color: black;">能够</span>输出多种文件类型。你<span style="color: black;">能够</span>在<span style="color: black;">这儿</span>找到它。</span></span></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;"><span style="color: black;">针对</span>处理文本,<span style="color: black;">尤其</span>是PDF,到<span style="color: black;">日前</span>为止,Claude 2表现优秀。我<span style="color: black;">已然</span>把整本书粘贴到前一个Claude版本中,效果令人印象深刻,新模型更加强大。你<span style="color: black;">能够</span><span style="color: black;">瞧瞧</span>我以前的经历,以及<span style="color: black;">有些</span>可能有趣的提示,在<span style="color: black;">这儿</span>。我还给了它许多<span style="color: black;">繁杂</span>的学术<span style="color: black;">文案</span>,<span style="color: black;">需求</span>它总结结果,它做得很好!<span style="color: black;">乃至</span>更好的是,你<span style="color: black;">而后</span><span style="color: black;">能够</span><span style="color: black;">经过</span>后续问题进行审问材料:有什么证据支持这种<span style="color: black;">办法</span>?作者得出了什么结论?等等...</span></span></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;">需要担心的事项:这些系统仍会在有限的方式下进行虚构。<span style="color: black;">倘若</span>你想<span style="color: black;">保证</span>准确性,需要<span style="color: black;">检测</span>它们的结果。</span></span></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;">获取信息和学习</span></span></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;">最佳免费选项:</span><span style="color: black;">Bing</span></span></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;">付费选项:<span style="color: black;">一般</span>Bing是最佳<span style="color: black;">选取</span>。<span style="color: black;">针对</span>儿童,可<span style="color: black;">运用</span>可汗学院的</span><span style="color: black;">Khanmigo</span><span style="color: black;">,它<span style="color: black;">供给</span>由GPT-4驱动的良好AI辅导。</span></span></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;"><span style="color: black;">倘若</span>你要将AI用作搜索引擎,可能最好不要<span style="color: black;">这般</span>做。<span style="color: black;">因为</span>存在高度的虚构<span style="color: black;">危害</span>,且大<span style="color: black;">都数</span>AI<span style="color: black;">无</span>连接到互联网,<span style="color: black;">运用</span>Bing会是明智之选(我<span style="color: black;">意见</span>你<span style="color: black;">运用</span>Bing。谷歌的AI Bard<span style="color: black;">更易</span>虚构)。然而,<span style="color: black;">按照</span><span style="color: black;">近期</span>的试点<span style="color: black;">科研</span>,证据<span style="color: black;">显示</span>AI在<span style="color: black;">小心</span><span style="color: black;">运用</span>的<span style="color: black;">状况</span>下,常常比搜索<span style="color: black;">供给</span><span style="color: black;">更加多</span>有用的答案。<span style="color: black;">尤其</span>是在搜索引擎效果<span style="color: black;">不良</span>的<span style="color: black;">状况</span>下,<span style="color: black;">例如</span>技术支持、决定吃什么、获取<span style="color: black;">意见</span>等,Bing相较于谷歌是一个更好的起点。这是一个快速发展的<span style="color: black;">行业</span>,但<span style="color: black;">日前</span>你在这些用途上还是需要<span style="color: black;">小心</span>。你肯定不想陷入麻烦。</span></span></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;"><span style="color: black;">然则</span>更令人兴奋的是,AI<span style="color: black;">能够</span>用于<span style="color: black;">帮忙</span>教育,<span style="color: black;">包含</span><span style="color: black;">帮忙</span><span style="color: black;">咱们</span>学习。我<span style="color: black;">已然</span>写过AI<span style="color: black;">怎样</span>用于<span style="color: black;">教育</span>,并<span style="color: black;">帮忙</span>教师更<span style="color: black;">容易</span>地工作、使<span style="color: black;">她们</span>的课程更有效,但它<span style="color: black;">亦</span><span style="color: black;">能够</span>用于自主学习。你<span style="color: black;">能够</span>让AI解释概念,<span style="color: black;">得到</span>非常好的结果。这个提示是一个很好的自动导师,你<span style="color: black;">能够</span>在<span style="color: black;">这儿</span>找到一个直接链接来在ChatGPT中激活导师。<span style="color: black;">由于</span><span style="color: black;">咱们</span><span style="color: black;">晓得</span>AI可能在虚构,<span style="color: black;">因此</span>你明智的做法是(仔细!)用另一个<span style="color: black;">源自</span>双重<span style="color: black;">检测</span>任何关键数据。</span></span></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;">还有<span style="color: black;">更加多</span>?</span></span></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;">鉴于技术的<span style="color: black;">快速</span>进步,这些可能都是你将<span style="color: black;">运用</span>的最糟糕的AI工具,正如过去几个月的发展所示。我确信不久后我就需要制作新的指南。但请记住两个关键点仍然适用与AI:</span></span></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;">AI是一种工具,并不总是正确的工具。<span style="color: black;">思虑</span>到其缺点,仔细<span style="color: black;">思虑</span>它<span style="color: black;">是不是</span>适合你计划应用它的目的。</span></span></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;">存在许多你需要<span style="color: black;">重视</span>的伦理问题。AI<span style="color: black;">能够</span>用来侵犯版权、作<span style="color: black;">坏处</span>、窃取他人劳动成果或<span style="color: black;">操作</span>他人。而特定AI模型的构建方式及谁从其<span style="color: black;">运用</span>中受益<span style="color: black;">常常</span><span style="color: black;">亦</span>都是<span style="color: black;">繁杂</span>的问题,<span style="color: black;">日前</span>并不太清楚。<span style="color: black;">最后</span>,你有责任以道德的方式<span style="color: black;">运用</span>这些工具。</span></span></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;"><span style="color: black;">咱们</span>正<span style="color: black;">处在</span>一个快速演进的革命的<span style="color: black;">初期</span><span style="color: black;">周期</span>。你还有其他的<span style="color: black;">运用</span>体会要分享吗?请在评论中告诉我。</span></span></p>
    <div style="color: black; text-align: left; margin-bottom: 10px;"><img src="https://p26-sign.toutiaoimg.com/tos-cn-i-qvj2lq49k0/9e7eea64bcd94a1a845269de2aa04d14~noop.image?_iz=58558&amp;from=article.pc_detail&amp;lk3s=953192f4&amp;x-expires=1727527649&amp;x-signature=PLoIKhfRsUMT%2B7489sFb7S0Akz8%3D" style="width: 50%; margin-bottom: 20px;"></div>
    <p style="font-size: 16px; color: black; line-height: 40px; text-align: left; margin-bottom: 15px;"><span style="color: black;"><span style="color: black;">引用:How to Use AI to Do Stuff: An Opinionated Guide, ETHAN MOLLICK,oneusefulthing.org</span></span></p>




j8typz 发表于 2024-10-6 08:12:05

系统提示我验证码错误1500次 \~゛,

m5k1umn 发表于 2024-11-13 09:01:28

软文发布论坛开幕式圆满成功。 http://www.fok120.com

4lqedz 发表于 5 天前

我完全同意你的看法,期待我们能深入探讨这个问题。
页: [1]
查看完整版本: 怎么样运用人工智能来办事:AI实用指南