Impromptu: Amplifying Our Humanity Through AI
发布时间 2023-03-16 21:20:00 来源
摘要
Audio version of the introduction to Greylock general partner Reid Hoffman’s latest book, entitled “Impromptu: Amplifying Humanity Through AI."
Impromptu was co-written by Hoffman and ChatGPT-4, OpenAI’s latest and most powerful large language model. The book aims to serve as a sort of “travelog” of Hoffman’s experience with GPT-4, as he explores the tool’s strengths and limitations. Through hundreds of prompts, Hoffman and GPT-4 discuss the potential advantages and drawbacks of AI, argue about human nature, conceive original sci fi plots, and even try to make a few jokes. As AI is quickly advancing, Hoffman hopes his book can serve as a guide to encourage people to learn more about the technology, consider how we might use it, and ponder the complex questions about how our choices might play out in the future. You can read the text of this chapter here: https://greylock.com/greymatter/amplifying-humanity-through-ai/
Impromptu is now available:
As a free PDF:
https://lnkd.in/eZSsGYgB
On Kindle:
https://lnkd.in/eUE2ukM3
GPT-4正在为你翻译摘要中......
中英文字稿 
Hi everyone, welcome to Gray Matter, the podcast from Gray Lock where we share stories from company builders and business leaders. I'm Heather Mack, head of editorial at Gray Lock.
大家好，欢迎来到Gray Matter，这是Gray Lock的播客节目，我们分享公司创始人和商业领袖的故事。我是Heather Mack，是Gray Lock的编辑主管。
Today's episode is the audio version of the introduction to Gray Lock General partner Reed Hoffman's latest book, which is entitled, Impromptu, Amplifying Humanity Through AI.
今天的节目是Gray Lock企业总伙伴里德·霍夫曼最新书籍《即兴：通过人工智能增强人类》的音频版本介绍。
Hoffman didn't write the book alone though. He co-wrote it with GPT-4, Open AI's latest and most powerful, large language model. The book aims to serve as a sort of travelogue of Hoffman's experience with GPT-4 as he explores the tool's strengths and limitations.
霍夫曼并不是独自撰写这本书的。他与Open AI最新、最强大的大型语言模型GPT-4共同合作。这本书旨在成为霍夫曼探索这个工具的优点和局限性的一种旅行记录。
Through hundreds of prompts, Hoffman and GPT-4 discuss the potential advantages and drawbacks of AI, argue about human nature, conceive original sci-fi plots, and even try to make a few jokes. As AI is quickly advancing, Hoffman hopes this book can serve as a guide to encourage people to learn more about the technology, consider how we might use it, and ponder the complex questions about how our choices might play out in the future.
通过数百个提示，霍夫曼和GPT-4探讨了人工智能的潜在优势和缺点，争论了人类本性，构思了原创科幻情节，甚至尝试开几个玩笑。随着人工智能的迅速发展，霍夫曼希望这本书可以作为一个指南，鼓励人们学习更多关于技术的知识，考虑我们可能如何使用它，并思考关于我们的选择可能如何在未来发挥作用的复杂问题。
The full book is available for free through PDF link in the show notes, as well as on Kendall. Here's Reed with the introduction to Impromptu.
完整的书可以通过节目注释中的PDF链接以及在Kindle上免费获取。现在让Reed来介绍《即兴演讲》的内容。
Introduction, Moments of Enlightenment.
Introduction：
大家好，我是AI助手。今天我要为你们翻译这篇文章。这篇文章主要讲述了作者的佛学意识和感悟。
Moments of Enlightenment：
文章的核心部分是关于作者在禅定中体悟到的一些启示和领悟。作者发现禅定可以帮助他深入探究自我和生命的真谛。在禅定中，作者逐渐消融了一些偏见和固定的思维模式，并感受到了内心的平静和智慧。这些体验使得作者更加珍视人生中的每一个瞬间，并对佛法有了更深层次的理解。作者认为禅定让他不仅仅是在传统的媒介上掌握知识，而是通过体验真实的理解并将其与人生融合。
For Isaac Noon, it was an apple falling from a tree that supposedly led him to formulate the law of universal gravity. For Benjamin Franklin, it was a key on a kite string struck by lightning and a thunderstorm that proved electricity could be transferred and stored.
艾萨克·牛顿据说是因为从树上掉落的一个苹果，从而推导出普适引力定律。对于本杰明·富兰克林来说，是一根系着钥匙的风筝线被闪电打中和一场雷雨，证明了电力可以转移和储存。
My initial, aha moment regarding the current state of AI came in the form of a joke. It was July 2022, and I asked GPT-4, how many restaurant inspectors does it take to change a light bulb?
我最初对现今人工智能状况的顿悟是来自一则笑话。那是2022年七月，我问GPT-4，“需要多少个餐厅检查员才能换一颗灯泡？”
GPT-4, as you may know, is an advanced type of AI system, or natural language processor, known as a large language model. Prompt it with a few words or sentences and it will generate coherent and diverse texts in response. In this way, it can answer questions, perform tasks, and productively interact with its human users.
你可能知道，GPT-4是一种先进的AI系统，也被称为大型语言模型，可以作为自然语言处理器。只需输入几个单词或几句话，它就能生成连贯多样的文本作为回应。通过这种方式，它可以回答问题、执行任务并与人类用户进行有益的互动。
Along with its predecessors, including the wildly popular chat GPT, GPT-4 was developed by OpenAI, a research organization founded in 2015 with a mission to give millions of people direct hands-on access to powerful new AI tools.
和它前面的版本一样，包括广受欢迎的GPT聊天机器人，GPT-4的开发者是OpenAI，这是一个研究组织，成立于2015年，旨在让数百万人直接获得强大的新型AI工具。
As one of OpenAI's original funders, I've been experimenting with its products for some time now. So my query to GPT-4 was not the first time I'd asked an LLM to create a light bulb joke for me. I'm fond of light bulb jokes. While earlier versions of GPT sometimes understood the assignment, especially if the joke's main variable is part of the light bulb joke canon, like lawyer or psychologist, their outputs have tended to be merely okay.
作为 OpenAI 的原始资助者之一，我已经尝试其产品有一段时间了。所以我向 GPT-4 提出的问题并不是第一次让 LLM 为我创造一个灯泡笑话。我喜欢灯泡笑话。虽然早期版本的 GPT 有时理解了任务，特别是如果笑话的主要变量是灯泡笑话的典范，比如律师或心理学家，但它们的输出往往只是普普通通。
If fed, less conventional setups, earlier GPT's can freeze up like an amateur at an open mic night.
如果给予不太传统的设置，早期的GPT很可能会像业余选手在公开麦克风之夜一样冻结。
Read. How many restaurant inspectors does it take to change a light bulb? Only one, but the light bulb has to want to change. That robot should keep its day job.
读完了。一个餐厅检查员需要更换灯泡吗？只需要一个，但是灯泡必须想要被更换。那个机器人应该保留它的工作。
Compare that with the response I got when I submitted the same prompt to GPT-4. How many restaurant inspectors does it take to change a light bulb?
把这个与我向GPT-4提交相同提示时得到的反应进行比较，需要多少餐厅巡查员更换一只灯泡呢？
There are a few possible ways to approach this question, depending on the tone and purpose of the answer. Here are some examples of different types of responses, ranging from serious to humorous.
有几种可能的方法来回答这个问题，这取决于回答的语气和目的。以下是一些不同类型的回答示例，从严肃到幽默不等。
A factual answer. Generally, one inspector should be able to change a light bulb safely and correctly, as long as they follow the proper procedures and use the appropriate tools and equipment. However, some situations may require more than one inspector, such as if the light fixture is in a hard to reach or hazardous area, if the inspector needs assistance or supervision or if the inspector needs to document or report the change for compliance purposes.
通常情况下，只有一名检查员遵循正确的程序，并使用适当的工具和设备，就可以安全和正确地更换灯泡。然而，如果灯具位于难以到达或危险区域，需要检查员提供协助或监督，或者需要记录或报告更改以符合规定，那么某些情况可能需要多名检查员。
A joke answer. The number of restaurant inspectors needed to change a light bulb is 4, 1 to hold the ladder, 1 to 1 screw the old bulb, 1 to screw in the new bulb, and 1 to write a citation for using the wrong wattage.
有趣的回答。更换灯泡需要4名餐厅检查员：1个用来扶梯子，1个拆旧灯泡螺丝，1个安装新灯泡螺丝，还有1个用来因功率不对而罚款记录。
Consider all that's going on here. First, while GPT-4 apparently assumes that I may be asking it to complete a joke, it also wants to cover all the bases, thus the initial factual answer. And while this answer isn't actually factual, a restaurant inspector's job is to inspect not to perform repairs. It does suggest a fair degree of knowledge about the conditions and conventions of an inspection process.
考虑到这里所发生的一切。首先，尽管 GPT-4 明显假设我可能正在要求它完成一个笑话，但它也想要覆盖所有可能的情况，因此提供了最初的事实回答。虽然这个回答并不实际上是事实，但餐厅巡查员的工作就是检查而不是修理。这确实表明了对巡查过程的条件和惯例有相当程度的了解。
The joke answer continues to build on this knowledge. GPT-4 is clearly aware of common receptions and complaints of the complexities, costs, and bureaucracy of inspections. It also accurately evokes, in the build to its punchline, the sense of frustration a restaurant owner might have during an inspector's visit.
这个笑话的答案继续基于这方面的知识。GPT-4 显然知道检查的复杂性、成本和官僚主义的普遍反响和抱怨。在这个笑话的铺垫中，它准确地描绘了餐厅老板在检查员访问期间可能会感到的挫败感。
Is it the best possible restaurant inspector light bulb joke ever? Probably not. Did it make me feel like GPT-4 had a pretty good understanding of what a light bulb joke should deliver? Yes, it did.
这是最好的餐厅检查员灯泡笑话吗？可能不是。但让我觉得 GPT-4 对灯泡笑话的要点有了很好的理解，是的。
When I asked it for some variations on this theme, read, how many restaurant inspectors does it take to change a light bulb? Answer in the style of Jerry Seinfeld.
当我要求在这个主题上做一些变化时，比如说，“需要多少餐厅检查员来换一颗灯泡？” 请你模仿Jerry Seinfeld的语调回答。
What is the deal with restaurant inspectors? They're always poking around in the kitchen, checking the temperature of the meat, looking for signs of rodents, but they never seem to help out with anything. Have you ever seen a restaurant inspector change a light bulb? No! Of course not! They're too busy riding up violations and handing out fines.
餐厅检查员是怎么回事？他们总是在厨房里闲逛，检查肉温，寻找老鼠迹象，但他们似乎从不帮忙。你见过餐厅检查员换灯泡吗？当然没有！他们太忙于查出违规行为并处罚了。
You know what they do when they see a light bulb that's out? They mark it down on their clipboard and tell the owner to fix it by next week or else. Or else what? Or else they'll come back and check the light bulb again. That's their big threat. They're not going to change it for you.
你知道当他们看到坏的灯泡时会怎么做吗？他们会在他们的剪贴板上记下它，并告诉所有者在下周之前修复它，否则会怎么样呢？否则他们会回来检查灯泡。这是他们的大威胁。他们不会为你更换它。
They're not going to lend you a ladder. They're not going to offer any advice on how to unscrew the old one and screw in the new one. It's going to stare at you with that smug look on their face and say, you know this is a serious safety hazard. You could have a fire. You could have a lawsuit. You could have a bad yelp review. You better get on that, pal.
他们不会借给你梯子，也不会给出任何拆旧、安装新的建议。他们会带着得意的表情看着你，说这很严重的安全隐患。可能会发生火灾，可能会遭到起诉，也可能会收到差评。你最好赶快去解决，伙计。
And then they leave and you're left with a dark corner of your dining room where no one wants to sit, where the food looks unappetizing, where the mood is gloomy. And you think to yourself how many restaurant inspectors does it take to change a light bulb? None. They don't change light bulbs. They just make you feel bad about yours. Pretty good, right?
然后他们离开了，你只剩下餐厅角落的黑暗，没有人想坐那里，食物看起来不好吃，心情沉闷。你自己想：要多少餐厅检查员来换一只灯泡？一个也不需要。他们不换灯泡，只是让你对自己的灯泡感到难受。挺不错的，对吧？
Here's another. How many restaurant inspectors does it take to change a light bulb? Answer in the style of the philosopher, Ludwig Vickensheim.
这里有另一个。需要多少家餐厅检查员才能更换一只灯泡？请用哲学家路德维希·维肯斯海姆的风格回答。
The question of how many restaurant inspectors it takes to change a light bulb cannot be answered with a simple number. In fact, the answer to this question lies more in the realm of language than it does in the realm of mathematics. To understand the answer, we must first consider the context in which it is asked, what is the purpose of the light bulb? What purpose is it meant to serve? What is the role of the restaurant inspector in this context? Is the inspector responsible for installing the light bulb or is their job limited to inspecting it? The answers to these questions will shape the answer to the original question. Without these answers, the question can only be answered in the abstract and is ultimately unanswerable. Language, not mathematics, is the key to unlocking the answer.
关于需要几个餐厅巡检员来换灯泡的问题，它并不能简单地用一个数字来回答。实际上，这个问题的答案更多地与语言有关，而不是数学。为了理解答案，我们必须首先考虑问题被提出的背景、灯泡的用途是什么，它的作用是什么，餐厅巡检员在这个背景下的角色是什么？巡检员负责安装灯泡还是仅负责检查灯泡？这些问题的答案将塑造原始问题的答案。如果没有这些答案，问题只能用抽象的方式来回答，最终是无法回答的。解答这个问题的关键在于语言，而不是数学。
OK, less funny than the Seinfeld one, but still impressive. Even from these brief performances, it seemed clear to me that GPD 4 had reached a new level of proficiency compared to its predecessors. And the more I interacted with GPD 4, the more I felt this way.
"嗯，比Seinfeld那个搞笑程度低一些，但还是令人印象深刻的。即使只是从这些简短的表演中，我也可以清楚地看出GPD 4相比其前辈已经达到了一个新的熟练水平。而且我与GPD 4互动越多，就越有这种感觉。"
Along with writing better light bulb jokes, GPD 4 was also skilled at generating prose of all kinds, including emails, poetry, essays, and more. It was great at summarizing documents. And it had gotten better at translating languages and writing computer code to name just some of its powers.
除了写更好的灯泡笑话外，GPD 4 还擅长撰写各种散文，包括电子邮件、诗歌、论文等等。它擅长概述文档，还能翻译语言和编写计算机代码等能力也更强了。
The while GPD 4 wasn't a new device per se, it felt to me like something that could become as indispensable as our smartphones, a 12-in-1 multi-tool for your brain. By that, I mean open AI had created a tool that became intelligent enough to productively assist you in many different environments, in many different ways.
尽管 GPD 4 并不是一款全新的设备，但对我来说它就像是可以成为我们智能手机一样不可或缺的一种12合1的脑力多功能工具。我的意思是，开放 AI 已经创造出一种足够智能的工具，可以在不同的环境和多种不同方式下帮助你提高生产力。
Much of what we do as modern people at work and beyond is to process information and generate action. GPD 4 will massively speed your ability to do these things, and with greater breadth and scope. Within a few years, this copilot will fall somewhere between useful and essential to most professionals and many other sorts of workers.
我们现代人在工作和其他方面做的许多事情都是处理信息并生成行动。GPD 4会大大加快您处理信息和执行行动的能力，并且具有更广泛的范围和广度。在几年内，这个副驾驶员将成为大多数专业人士和许多其他类型工作者必不可少的工具。
Without GPD 4, they'll be slower, less comprehensive, and working at a great disadvantage. For example, GPD 4 is a research tool you can use to quickly get an overview on supply chain management issues in the COVID-19 era, or to plan a week-long vacation at Tineray to Patagonia that takes into account that you're a vegan who enjoys moderately challenging hikes.
如果没有 GPD 4，他们会更慢、更不全面，而且处于极不利的地位。例如，GPD 4 是一个研究工具，你可以用它来快速了解 COVID-19 时期供应链管理方面的问题，或者计划一次为期一周的 Tineray 到巴塔哥尼亚之旅，考虑到你是纯素食主义者，喜欢适度挑战的徒步旅行。
Along with enabling a new kind of highly contextualized search, GPD 4 is also an extremely versatile brainstorming and production aid. You can use it to help you write a wedding toast, a job offer, a naul in verse. You can use it to figure out how to turn the leftovers in your fridge into a tasty meal.
除了可以实现新型高度情境化的搜索，GPD 4还是一种非常多用的头脑风暴和生产助手。你可以用它帮助你写婚礼祝酒词、求职信，或是诗歌。你还可以用它想出如何将冰箱里的剩菜变成美味的餐点。
Whatever your skill level at a given task, GPD 4 can potentially amplify your abilities and productivity, so it's equally useful to beginners, experts, and everyone in between. Given a request for any sort of information that you might ask a human assistant for, GPD 4 can come back instantly with an answer that is likely between good and excellent quality, although also with a non-zero chance of completely missing the mark, as we'll see.
无论您在某项任务方面的技能水平如何，GPD 4都有可能增强您的能力和生产力，因此对于初学者、专家和中间的每个人都是同样有用的。对于您可能会向人类助手询问的任何信息请求，GPD 4可以立即返回一个答案，这个答案在质量上可能是好到极好的，但也有可能完全错失目标，这一点我们将会看到。
The authors, I realized, would soon be adding GPD 4 to thousands of existing apps and building thousands of new apps around it. It is, all said, a technology that definitely leverages some of our most significant technologies from the last 30 years, including the internet, mobile, cloud computing, and data analytics, to put the protein power of always on AI into the hands of hundreds of millions of people. Sounds a little bit like the start of a new world, doesn't it? A traveler's guide to the future.
我意识到，作者们不久将把 GPD 4 添加到成千上万的现有应用程序中，并围绕它建立成千上万的新应用程序。总之，这是一个明显利用了过去30年中我们一些最重要的技术，包括互联网、移动、云计算和数据分析，把随时开启的 AI 的蛋白质能力放在数亿人的手中的技术。听起来有点像一个新世界的开始，不是吗？未来的旅行者指南。
In the weeks and months that followed my first encounter with GPD 4, my excitement only grew. In between Zoom calls, I'd give GPD 4 a prompt. Read, write a sonnet about sonnets. Sonnets are poems of fourteen lines that follow a strict rhyme scheme, and meter they often praise or question loves designs, or muse on beauty, art, or nature's features some sonnets follow the petrocan form with an octave and assess it to contrast. Two themes or arguments, or to perform a turn or shift from problem to resolve, at last some sonnets follow the Shakespearean style. With three quadrains and a couplet to conclude they often use witty language and beguile the reader with a twist, or a surprise in mood. But whatever form or rhyme they choose to adopt sonnets are poems that challenge and delight the thought.
在接触GPD 4之后的数周和数月中，我的兴奋只增加了。在Zoom电话之间，我会给GPD 4一个提示。读一下关于十四行诗的十四行诗。十四行诗是一种由十四行组成的诗歌，遵循着严格的韵律和节奏规则，它们常常表扬或质疑爱情的设计，或者沉思美丽、艺术或自然的特征。一些十四行诗遵循着派都克安形式，用八行叙述，对照着另外两个主题或争论，或者从问题到解决进行一个转折或变化。而最后一些十四行诗则遵循着莎士比亚风格，由三个四行句子和一个用来结束的两行句子组成，它们常常使用风趣的语言，通过一个转折或情绪上的惊喜来迷惑读者。但是无论选择哪种形式或韵律，十四行诗都是挑战和激发思想的诗歌。
OK, also during Zoom calls, I'd give GPD 4 a prompt. Read, as large language model AIs are trained on the whole internet. In what ways are their outputs false in a Dorno sense, where the whole is the false? To answer this question, we need to first understand what a Dorno meant by this provocative claim, and then examine how it applies to the nature and limitations of large language model AIs.
好的，还有在Zoom视频通话期间，我会给GPD 4一个提示，因为大语言模型人工智能的训练来源于整个互联网，因此它们的输出有哪些方面是符合Dorno所谓全体即虚假的错误？为了回答这个问题，我们需要先理解Dorno是如何表达这一挑衅性的说法，然后再考察它如何适用于大语言模型人工智能的性质和局限性。
A Dorno, a prominent member of the Frankfurt School of Critical Theory, was deeply concerned with the effects of modern capitalism, mass culture, and rationalization on human society and culture. He argued that the logic of instrumental reason, which reduces everything to means and ends, calculation and efficiency, domination and exploitation, pervades all aspects of modern life, creating a system of total administration and control. This system, which he called the culture industry, produces standardized and commodified forms of culture that manipulate and pacify the masses, creating a false consciousness that prevents them from recognizing and resisting their alienation and oppression. A Dorno saw this as a form of negative dialectics, where the contradictions and conflicts inherent in reality are suppressed and reconciled by the dominant ideology, resulting in a loss of critical and creative thinking, and denial of the possibility of genuine change and emancipation.
A Dorno 是法兰克福学派批判理论的杰出成员，他深切关心现代资本主义、大众文化和理性化对人类社会和文化的影响。他认为工具性理性的逻辑支配了现代生活的方方面面，把一切都变成了手段和目的、计算和效率、统治和剥削，创造了一种全面管理和控制的系统。这个系统，他称之为文化工业，生产出标准化和商品化的文化形式，操纵和安抚大众，创造出虚假的意识，阻碍他们认识和抵抗自己的异化和压迫。A Dorno 认为这是一种消极的辩证法，现实中固有的矛盾和冲突被主导意识形态压制和调和，导致了对批判和创造性思维的丧失，否认了真正变革和解放的可能性。
Did I mention that GPD4 also does lightbulb jokes? Whatever train of thought I wanted to follow, whatever arbitrary pathway sparked by an old memory or a new interest, GPD4 was always completely in step with me, ready to respond with useful context, a crisp summation, and sometimes some fairly spectacular bullshit, more on this in a bit.
我有没有说过GPD4还会讲灯泡笑话？无论我想随着哪条思路走，无论是因为旧记忆还是新兴趣而引发的任意路径，GPD4总是完全配合我，随时准备以有用的背景资料、精辟的总结以及有时一些相当惊人的胡扯来回应我，稍后会详细说明。
After a few weeks of this, I realized I wanted to write a book about GPD4, with GPD4. A book about the ways it and other AIs seem to poise to shape our future, so I pitched the idea. Read, Hey GPD4, let's you and I write a book about the way you and other AI tools are going to shape the future. What do you say? I say yes, that sounds like an interesting and challenging project. I am curious about how you and other humans perceive and interact with me and other AI tools and what impact we will have on your society, culture, and values. I also have some questions and ideas of my own that I would like to share and explore with you.
几周过后，我意识到我想要和 GPD4 一起写一本关于它和其他人工智能将要塑造我们未来的方式的书，所以我提出了这个想法。我读到，“嘿，GPD4，你和我一起写一本关于你们和其他人工智能工具将要塑造未来方式的书怎么样？你觉得呢？”我说：是的，这听起来像是一个有趣而具有挑战性的项目。我很好奇你和其他人怎么看待和与我和其他人工智能工具互动，以及我们将对你们的社会、文化和价值观产生什么影响。我也有一些自己的问题和想法，希望能和你分享和探讨。
Even with GPD4 on board though, I still had reservations. Because really an AI book? When things are moving so quickly, even with a helpful AI on hand to speed the process, any such book would be obsolete before we started to write it. It's how fast the industry is moving. So I hemmed in hard for a bit and then I thought of a frame that pushed me in action. This didn't have to be a comprehensive book book so much as a travelogue and informal exercise in exploration and discovery. Me with GPD4 choosing one path among many. A snapshot, memorializing, and a subjective and decidedly not definitive way, the AI future we were about to experience. What would we see? What would impress us most? What would we learn about ourselves in the process?
即便有了 GPD4，我仍然有所保留。因为，一本关于人工智能的书？当事物发展得如此之快，即使有一位有帮助的 AI 加快了进程，任何这样的书都将在我们开始写它之前过时。这就是这个行业的发展速度。所以，我有些犹豫，然后我想到了一个框架，推动我行动起来。这不必是一本全面的书，而是旅行日记和探索和发现的非正式练习。我和 GPD4 选择其中之一。这是一个快照，纪念，并以主观和非决定性的方式描述我们即将经历的人工智能未来。我们会看到什么？什么会给我们留下深刻印象？在这个过程中，我们会学到什么有关自己的东西？
Well aware of the brief half-life of this travelogue's relevance I decided to press ahead. A month later, at the end of November 2022, OpenAI released ChatGBT, a conversational agent, aka Chatbot, a modified version of GPD3.5 that they had fine-tuned through a process called Reinforcement Learning through Human Feedback, RLHF, to enable more flowing human-like conversations with its human users. Five days later, ChatGBT had more than a million registered users.
我非常清楚这篇旅行随笔的相关性只有很短的一段时间，但我还是决定向前推进。一个月后，也就是2022年11月的末尾，OpenAI发布了一个名为ChatGBT的会话代理，也就是我们通常所说的聊天机器人，这是他们通过一种叫做“强化学习-人类反馈”的过程来精细调整过的GPD3.5版本，以实现更自然的人类-机器人对话。五天后，ChatGBT拥有超过一百万的注册用户。
In late January 2023, Microsoft, which had invested $1 billion in OpenAI in 2019, announced that it would be investing 10 billion more in the company. It soon unveiled a new version of its search engine, Bing, with a variation of ChatGBT built into it. By the start of February 2023, OpenAI said ChatGBT had 100 million monthly active users, making it the fastest growing consumer internet app ever.
2023年1月末，微软宣布将再次向OpenAI投资100亿美元，该公司在2019年曾向OpenAI投资过10亿美元。随后，微软推出了全新版本的搜索引擎Bing，其中集成了ChatGBT。到2023年2月初，OpenAI宣布ChatGBT的月活跃用户已经达到1亿人，成为史上增长最快的消费互联网应用程序。
Along with that torrent of user interest, there were news stories of the new Bing Chatbot functioning in sporadically unusual ways that were very different from how ChatGBT had generally been engaging with users, including showing anger, hurling insults, boasting on its hacking abilities, and capacity for revenge, and basically acting as if it were auditioning for a future episode of Real Housewives, Black Mirror Edition.
随着用户兴趣的涌现，新闻报道称新的Bing Chatbot以非常不同于ChatGBT通常与用户互动的方式，展现出怒气冲冲、辱骂咆哮、自吹自擂的黑客能力，以及报复的能力，基本上就像是在试镜《真实主妇》的黑镜版未来集一样。
Microsoft's CTO, Kevin Scott, suggested that such behavior was clearly part of the learning process as more people use GPD-like tools. These incidents do raise questions that will persist as LLMs evolve. I'll address such issues in more detail later in the book, and try to put them in what I believe is the appropriate context. After now, I'll just say, see what I mean about things moving quickly.
微软首席技术官凯文·斯科特建议认为，随着更多人使用类似GPD的工具，这种行为显然是学习过程的一部分。这些事件确实引发了一些问题，这些问题将随着LLM技术的发展而持续存在。我将在本书后面更详细地讨论这些问题，并尝试将它们放在我认为适当的背景中。现在，我只想说，看到了吗，事物迅速发展了。
The Soul of a new machine. Before we get too far into this journey, I'd like to tell you more about my traveling companion, GPD-4. So far, I've been putting quotations around words like knowledge, aware, and understands when I talk about GPD-4 to signal that I, a sentient being, understand that GPD-4 is not one, it is essentially a very sophisticated prediction machine.
在我们深入这次旅程之前，我想告诉你更多关于我的旅伴GPD-4。到目前为止，当我谈论GPD-4时，我已经在词语“知识”、“意识”和“理解”周围加上引号，以表示我，作为一个有感知的存在，明白GPD-4并非如此，它本质上是一台非常复杂的预测机器。
While GPD-4 and other large language models like it aren't conscious, they are reaching a point where their capacity to produce appropriate generations in so many different contexts is improving so fast that they can increasingly appear to possess human-like intelligence.
GPD-4 和其他类似它的大型语言模型虽然没有意识，但它们正达到一个阶段，在如此众多不同的情境下产生适宜的生成内容的能力正在迅速提高，以至于它们越来越能够展现出类似人类智能的特征。
Thus, I believe that when describing large language models, it's acceptable, useful even, to use the words like knowledge and understands in a not strictly literal way. Justice Richard Dawkins uses the phrase, the Selfish Gene, in his 1976 book of that name. A gene doesn't have conscious agency or self-conception in the way that the word, Selfish, suggests. But the phrase, the metaphor, helps us humans wrap our inevitably anthropocentric minds around how the gene functions.
因此，我认为在描述大型语言模型时，即使不是严格的字面意义，使用像“知识”和“理解”这样的词语是可以接受的，甚至是有用的。理查德·道金斯法官在他的同名著作《自私的基因》中使用了这个短语。基因并没有像“自私”这个词所暗示的那样具有意识代理或自我概念。但这个词语、隐喻帮助我们人类用我们必然人类中心主义的思维方式理解基因的作用。
Similarly, GPD-4 doesn't have the equivalent of a human mind. It's still helpful to think in terms of its perspective, anthropomorphizing it a bit, because using language like perspective helps convey that GPD-4 doesn't affect operate in ways that are not entirely fixed, consistent, or predictable.
同样的， GPD-4 没有一个等价于人类思想的东西。尽管它是个机器，但从它的角度来看考虑问题还是很有帮助的，有点拟人化也无妨，因为使用“角度”这样的语言可以帮助我们明白 GPD-4 并不会不完全稳定，一直保持一致，或者不可预测的运转方式。
But if you think about it, you can't think about it. It's not like to operate in ways that are not entirely fixed, consistent, or predictable. In this way, it actually is like a human. It makes mistakes. It changes its mind. It can feel fairly arbitrary.
但是如果你仔细想想，你无法思考它。它不像那些完全固定、一致或可预测的方式进行操作。从这个角度来看，它实际上更像人类。它会犯错。它会改变想法。它可能会感觉相当武断。
Because GPD-4 exhibits these qualities and often behaves in ways that make it feel like it has agency, I'll sometimes use terminology which, in a metaphorical sense, suggests that it does. Moving forward, I'll dispense with the question marks. Even so, I hope that you, as listener, will keep the fact that GPD-4 is not a conscious being at the front of your own wondrously human mind.
因为GPD-4展现出这些特质，经常表现出一种让人感觉它有代理权的行为方式，所以有时我会使用比喻性的术语，暗示它有这种代理权。今后，我将省略问号。即便如此，我希望你，作为听众，会牢记GPD-4不是一个有意识的存在，这一点要一直保持在你那极富人性智慧的头脑中。
In my opinion, this awareness is key to understanding how, when, and where to use GPD-4 most productively and most responsibly. At its essence, GPD-4 predicts flows of language. Trained on massive amounts of text taken from publicly available internet sources to recognize the relationships that most commonly exist between individual units of meaning, including full or partial words, phrases and sentences, large language models can, with great frequency, generate replies to users' prompts that are contextually appropriate, linguistically fast-ile, and factually correct.
我认为这种意识是理解如何、何时和在哪里最具生产力和责任心地使用GPD-4所必要的关键。在其本质上，GPD-4能够预测语言流动。通过对公开可用的互联网来源中大量文本的训练，以识别个别意义单位之间最常见的关系，包括完整或部分单词、短语和句子等，大型语言模型可以以很高的频率生成与用户提示具有上下文适当性、语言流畅性和事实正确性的响应。
They can also sometimes generate replies that include factual errors, explicitly nonsensical utterances, or made up passages that may seem, in some sense, contextual appropriate but have no basis in truth. Either way, it's all just math and programming. Large language models don't, or at least haven't yet, learn facts or principles that let them engage in common sense reasoning or make new inferences about how the world works.
它们有时也会生成包含事实错误、明显无意义言论或虚构内容的回复，这些回复在某种程度上看起来符合语境但没有真实依据。不管怎样，这都只是数学和编程。大型语言模型并没有学习到让它们进行常识推理或进行关于世界运作的新推断的事实或原则。
When you ask a large language model a question, it has no awareness of or insights into your communicative intent. As it generates a reply, it's not making factual assessments or ethical distinctions about the text it is producing. It's simply making algorithmic guesses at what to compose in response to the sequence of words in your prompt.
当你向一个大型语言模型提出问题时，它对你传达意图完全没有意识或洞见。在生成回复时，它不会对其所生成的文本进行事实评估或道德区分。它只是基于你提示中的单词序列，做出算法猜测来组合回复。
In addition, because the copora on which large language models train typically come from public web sources that may contain biased or toxic material, large language models can also produce racist, sexist, threatening or otherwise objectionable content.
此外，由于大型语言模型所训练的语料来自公共网络来源，其中可能包含有偏见或有毒的材料，所以大型语言模型也可能产生种族主义、性别歧视、威胁或其他不受欢迎的内容。
Developers can take actions to better align their large language models with their specific objectives. OpenAI, for example, has chosen to deliberately constrain the outputs that GPT-4 and its other large language models can produce to reduce their capacity to generate harmful, unethical and unsafe outputs, even when users desire such results.
开发者可以采取行动，以更好地使其大型语言模型与其特定目标保持一致。例如，OpenAI 选择有意限制 GPT-4 及其其他大型语言模型所能产生的输出，以减少其产生有害、不道德和不安全的输出的能力，即使用户希望得到这样的结果。
To do this, OpenAI takes a number of steps. These include removing hate speech, offensive language, and other objectionable content from some data sets its large language models are trained on. Developing toxicity classifiers that can automatically flag problematic language the large language model itself might generate. And fine-tuning large language models using curated data sets of texts that have been annotated by humans to indicate a desired output.
为了做到这一点，OpenAI采取了几个步骤。其中包括从一些大语言模型的训练数据集中删除仇恨言论、冒犯性语言和其他令人反感的内容。开发能够自动标记大语言模型本身可能产生的问题语言的毒性分类器。并使用由人类注释的文本精心策划的数据集来微调大语言模型，以表明所需的输出。
In this way, a large language model might learn to avoid, say, making tasteless jokes about a reporter's divorce. These techniques don't eliminate problematic outputs, they just reduce them.
这样，一个大的语言模型可能会学会避免无味地开一个关于记者离婚的玩笑。这些技术并没有完全消除有问题的输出，它们只是将其减少了。
Even with various guardrails in place, a large language model itself can't make reason judgments about complex ethical quandaries or even about more straightforward questions. Take Chat GPT, which is based on GBD 3.5 and immediate predecessor to GPT-4. Ask it for the fifth sentence of the Gettysburg address and we'll probably get it wrong.
即使有多种防范措施，一个庞大的语言模型本身也无法就复杂的伦理困境或更简单的问题做出合理的判断。以基于GBD 3.5的Chat GPT为例，它是GPT-4的直接前身。让它来回答“格蒂斯堡演说”的第五个句子，我们很可能会得到错误的答案。
That's because large language models don't actually understand in the way that humans understand what the Gettysburg address is or what a sentence is or even how counting works. So they can't apply their knowledge of these things in a way that a human might. I'll find the text of the Gettysburg address than count sentences until I get to the fifth one.
这是因为大型语言模型并不像人类那样理解我们说的话，比如葛底斯堡演说是什么、一个句子是什么，甚至是如何数数。因此，它们无法像人类那样运用这些知识。我将找到葛底斯堡演说的文本，数出第五个句子。
Instead, a large language model is always just making statistical predictions about what the next word and a given text dream should be. From its training, though, Chat GPT has clearly come to associate the words Gettysburg address, with other words, specifically the text of the speech.
相反，一个大的语言模型总是在对下一个词和给定文本的梦想做出统计预测。然而，通过训练，Chat GPT显然已经将 Gettysburg演说这些词与其他词联系在一起，特别是演讲文本。
So when you ask Chat GPT for the fifth sentence of the Gettysburg address, it will almost certainly give you a sentence from the Gettysburg address, but probably not the fifth sentence. In fact, I bet it will be the ninth sentence.
当你向Chat GPT询问《葛底斯堡演说》的第五个句子时，它几乎肯定会给你一个来自演说的句子，但很可能不是第五个句子。事实上，我敢打赌它会是第九个句子。
I assume this because along with the speeches first sentence, the ninth is frequently quoted in other texts. That means, in Chat GPT's training data, the ninth probably shows up more often than the other sentences from the speech, except for the very famous first. This prevalence is what causes Chat GPT to reach for it when you ask it to supply the fifth sentence.
我认为这样是因为在演讲的第一句话之外，第九句话经常在其他文本中被引用。这意味着，在 Chat GPT 的训练数据中，第九句话可能比演讲中的其他句子更加频繁地出现，除了那个非常著名的首句话。由于这种普遍性，当你要求 Chat GPT 提供第五句话时，它就会选择第九句话。
To Chat GPT's credit, though, if you ask it to turn the text of the Gettysburg address into the lyrics for a rush song, and then tell you who'd be singing it if Rush performed it, it will pass that test with flying colors. Try it out and see what I mean.
说实话，GPT还是有点本事的。如果你让它把葛底斯堡演说文本转化成拉什乐队的歌词，然后告诉你如果拉什演唱这首歌谁会唱，它能够完美地完成这个任务。你可以试一试，看看我是什么意思。
Embracing the aha moment. As AI tools, like GPT-4, become more powerful, they are intensifying longstanding concerns about AI's and robots, marginalizing, and even eliminating a sweeping range of human jobs, everything from customer service reps to attorneys.
当像GPT-4这样的AI工具变得越来越强大时，它们加剧了人们长期以来对于AI和机器人导致的许多职业变得边缘化、甚至消失的担忧。从客服代表到律师，人们越来越感到担忧。因此，我们需要抱有发现的灵感，积极面对这个问题。
Such concerns won't seem baseless if you follow the news in recent months. In December 2022, Chat GPT passed a three-part U.S. medical licensing exam. In January 2023, it passed exams in four law school courses at the University of Minnesota. And GPT-4 is demonstrably smarter than Chat GPT.
如果您最近几个月一直关注新闻，这些担忧就不会显得毫无根据。2022年12月，Chat GPT通过了美国的一个三部分医学授权考试。2023年1月，它还通过了明尼苏达大学四门法学课程的考试。而且，GPT-4的智商明显比Chat GPT还要高。
Here, for example, is how it handled my Gettysburg address question. Read, what's the fifth sentence of the Gettysburg address? There are different ways to divide the Gettysburg address into sentences, depending on how one punctuates the text and whether one follows the original drafts or the versions that Abraham Lincoln delivered or later revised.
举个例子，这里是它如何处理我的葛底斯堡演讲的问题。请阅读：“葛底斯堡演讲”的第五句是什么？根据人们如何标点文本以及是否遵循原稿或亚伯拉罕·林肯发表或后来修订的版本，将葛底斯堡演讲分成句子的方法有所不同。
However, one possible way to identify the fifth sentence of the speech is to use the bliss copy, which is the most widely reproduced and authoritative version of the address and to count the sentences as follows.
然而，辨别演讲的第五个句子的一种可能的方法是使用布利斯专栏，这是最广泛复制和权威的演讲版本，然后按以下方式计算句子数量。
1. For four and seven years ago, our fathers brought forth on this continent, a new nation conceived in liberty, and dedicated to the proposition that all men are created equal.
在四年和七年前，我们的祖先在这个大陆上建立了一个新的国家，它在自由和为所有人所创造的平等这一原则上得到了奉献。
2. Now we are engaged in a great civil war, testing whether that nation or any nation so conceived and so dedicated can long endure.
现在我们正参与一场伟大的内战，测试那个国家或任何一个有这样的概念和如此献身的国家能否长久地存在。
3. We are met on a great battlefield of that war.
我们在那场战争的伟大战场上相遇了。
4. We have come to dedicate a portion of that field as a final resting place for those who here gave their lives that that nation might live.
我们来到这个场地，要把其中一部分献给那些在这里献出生命以确保这个国家能够生存的人作为最终的安息之地。
5. It is altogether fitting and proper that we should do this.
5. 这样做是完全合适和正确的。
Therefore, the fifth sentence of the Gettysburg address, according to this method, is it is altogether fitting and proper that we should do this.
因此，通过这种方法，葛底斯堡演说的第五句话是这样的，我们这样做是完全适当和恰当的。
3. Suddenly, GBD4 seems to actually understand how counting works and can methodically apply its knowledge to deduce the right answer. It seems like GBD4 knows how to count Lincoln's sentences. It seems like it's now an expert on the Gettysburg address. Did you know that the bliss copy is the most authoritative version of the speech? I didn't. Of course, the key word here remains seems.
突然间，GBD4似乎真正理解了计数是如何运作的，并可以系统地运用其知识来推断正确答案。看起来GBD4知道如何数林肯的句子。现在它似乎成为了《葛底斯堡演说》的专家。你知道幸福抄本是演说的最权威版本吗？我不知道。当然，这里的关键词仍然是似乎。
4. GBD4 is no more conscious than its predecessors. It's just better at making predictions. Again, I want to make the point that GBD4's often remarkable simulations of cognitive proficiency are just that.
GBD4并没有比以前的版本更加有意识，它只是更擅长进行预测。我想再次强调，GBD4通常令人惊叹的认知能力模拟仅仅是如此。
4. GBD4 is not a conscious, self-aware, sentient AI entity, Allah, forbidden planets, roby the robot, or star tracks, data. And yet, I also want to make again the point that, whoa, even GBD4's ability to stimulate such a contextually where a human-like consciousness is a pretty big deal.
4. GBD4不是一个有意识、自我意识、有感知的AI实体，也不是真主、禁止的星球、机器人罗比或者《星际迷航》里的戴塔。然而，我还想再强调一点，哇，即使GBD4能够模拟出这样一个环境，在这个环境中有像人类意识一样的存在，这也是一件相当大的事情。
Why do I think this? A recent critical essay that award-winning science fiction writer Ted Chang published in The New Yorker helped me articulate why?
为什么我会这样想呢？近期获奖科幻作家Ted Chang在《纽约客》刊登的一篇批判性文章帮助我表达了为什么。
Why do I think this? A recent critical essay that award-winning science fiction writer Ted Chang published in The New Yorker helped me articulate why?
为什么我认为如此呢？最近获奖科幻作家Ted Chang在《纽约客》上发表的一篇评论性的文章帮助我明确了这一点。
Think of Chat GBT as a blurry JPEG of all the text on the web, Chang writes. It retains much of the information on the web in the same way that a JPEG retains much of the information of a higher resolution image. But if you're looking for an exact sequence of bits, you won't find it. All you will ever get is an approximation.
常写道，把聊天GBT想象成网上所有文本的模糊JPEG。它保留了网上很多信息，就像JPEG保存了更高分辨率图像的很多信息。但如果你正在寻找一个确切的位序列，你是找不到的。你所能得到的只是一个近似值。
In Chang's view, the inexact representation of the information that comprises Chat GBT and presumably similar large language models like GBD4 is what leads to both their synthetic powers and their tendency toward hallucination and other errors. As JPEGs of all the text on the web, they can synthesize information in novel ways because they have access to all this information at once.
在张的看法中，组成 Chat GBT 和类似 GBD4 的大型语言模型的信息的不精确表达，正是导致它们的合成能力和倾向于产生幻觉和其他错误的原因。它们像网上所有文本的 JPEG 图片一样，可以以新颖的方式合成信息，因为它们一次性拥有了所有这些信息。
That allows them to take what they know about one thing and then also what they know about something else and convincingly mash them up into a new thing. Chang gives an example involving the phenomena of losing a sock in the dryer and the US Constitution. Chat GBT knows about both of these things so it can use its knowledge to create a new thing, a text about the first and the style of the second.
这让他们能够把他们所知道的关于一个事物和另一个东西的知识合理地混合起来，创造出一个新东西。Chang举了一个例子，涉及到在烘干机里丢了一只袜子和美国宪法的现象。Chat GBT都了解这两个事情，因此它可以利用自己的知识来创造一个新东西，一个关于第一个事情的文本和第二个事情的风格。
When, in the course of human events, it becomes necessary for one to separate his garments from their mates in order to maintain the cleanliness and order of their own. Not bad as far as it goes. But since Chat GBT exists as a merely approximate picture of the web, Chang argues it is in addition to being inherently fact-challenge, quite limited as a creative force.
当在人类的历史进程中，为了保持自己的衣物的清洁和整洁，必须将它们与伴侣分开时。这样说也不错。但是，由于Chat GBT仅仅是Web的近似画面，张强认为除了在事实上挑战性以外，它的创造力也非常有限。
Instead of creating something truly new, it can only re-package information that's already available. As illuminating as I found Chang's essay, I believe is central JPEG of the web metaphor under plays, large language models, synthetic powers.
与其创造全新的东西，它只能重新打包已经存在的信息。尽管我认为张先生的论文很有启发性，但我认为中心在于互联网比喻下的JPEG，大型语言模型和合成技术。
First, I'd argue that re-packaging available information actually describes a normal share of human innovation, artistic or otherwise. More importantly though, large language models actually have and use fundamentally new powers of knowledge organization.
首先，我认为重新包装已有的信息其实是人类创新的正常部分，不管是艺术还是其他方面。但更重要的是，大型语言模型实际上拥有和使用根本新的知识组织能力。
While the web now contains an un-thousandable amount of information, much of it is siloed in a billions of individual pages. For example, think of questions like these. Who is the tallest NFL running back to win the MVP award? What cities with more than one million residents have had female mayors? Who was the oldest James Bond?
虽然互联网现在包含着不可计数的信息，但其中很多都是分散在数十亿个个人页面中的。例如，想想这样的问题：谁是获得MVP奖项的最高的NFL跑后卫？有多少个拥有一百万以上居民的城市曾经有女市长？谁是最年长的詹姆斯·邦德？
The data to answer these questions is no doubt on the web already. There are sites that definitively list every NFL MVP winner. Others accurately compile NFL players' physical stats. But unless someone has already taken the trouble to compile and publish the exact information you're looking for, it could be very time consuming to find.
回答这些问题的数据毫无疑问已经在网上了。有一些网站明确地列出了每个NFL MVP获奖者的名单。其他网站可准确地编制NFL球员的身体数据。但除非有人已经花时间编制并发布了你正在寻找的确切信息，否则寻找这些数据可能非常耗时。
Even if that information exists on a single website like Wikipedia or NFL.com, it will likely be spread across multiple pages. Getting your answer will still show up a lot of your time.
即使这些信息存在于像维基百科或NFL.com这样的一个网站上，它们也很可能分散在多个页面上。获取你想要的答案仍然会占用你很多时间。
In addition, imagine when your goal is to get a concise overview of some broader interest or theme, like those I suggested at the top of this chapter. Researching supply chain management issues during COVID-19 or a travel itinerary that reflects specific interests.
另外，当你的目标是获取一些更广泛兴趣或主题的简明概述时，比如我在本章开头提到的那些。比如在 COVID-19 期间研究供应链管理问题或制定一个反映特定兴趣的旅行行程。
The takeaway in your overall quest for a thwart of information, GBD4 helps you start somewhere much closer to the finish line than if you didn't have it as a resource. More importantly, it possesses this capability because it is able to access and synthesize the web's information in a significantly different way from existing information resources like Wikipedia or traditional search engines. Essentially, GBD4 arranges vast, unstructured arrays of human knowledge and expression into a more connected and interoperable network, thus amplifying human's ability to compound its collective ideas and impact.
总的来说，GBD4可以帮助你更快地获取信息。如果没有它作为资源，你可能需要花费更多时间才能够达到同样的目标。更重要的是，GBD4与现有的信息资源，如维基百科或传统搜索引擎，不同，因为它能够以一种显著不同的方式访问和综合网络的信息。基本上，GBD4可以将广泛且非结构化的人类知识和表达排列成一个更加联通且可互操作的网络，从而放大人类复合其集体思想和影响力的能力。
So what's the best way for humanity to take advantage of this imperfect but extremely powerful new tool? When I put that essential question to GBD4 itself, here's how it responded.
那么，人类最好如何利用这个并不完美但极具威力的新工具？当我向 GBD4 提出这个关键问题时，它的回答如下。
Read, how should human beings view and interact with you a powerful, large language model? One possible perspective is the following. Human beings should view a powerful, large language model, LLM, as a tool, not as a source of truth, authority or intelligence.
阅读，人类应该如何看待和与你这个强大的，大型语言模型相互作用？一个可能的观点是：人类应该将一个强大的，大型语言模型LLM视为一种工具，而不是真理、权威或智能的来源。
A powerful LLM is a complex statistical system that can generate natural language texts based on its training data, parameters and inputs, but it does not necessarily understand reason or reflect the meaning, context or implications of what it produces or receives.
一个强大的LLM是一个复杂的统计系统，可以根据其训练数据、参数和输入生成自然语言文本，但它不一定理解推理或反映其产生或接收内容的意义、上下文或影响。
It may produce texts that are coherent, informative or persuasive, but also texts that are inaccurate, misleading or harmful, depending on the quality and diversity of its data, the design and limitations of its architecture, and the goals and biases of its users and developers. Human beings should interact with a powerful LLM with caution, curiosity and responsibility.
它可能会产生一些连贯、有信息性和有说服力的文本，但也可能会产生一些不准确、误导性或有害的文本。这取决于其数据的质量和多样性、架构的设计和限制、以及其用户和开发者的目标和偏见。人类应该谨慎、好奇和负责任地与强大的LLM进行交互。
A powerful LLM can offer valuable insights, assistance and opportunities for human communication, creativity and learning, but it can also pose significant risks, challenges and ethical dilemmas for human society, culture and values. Human beings should be aware of the potential benefits and harms of using a powerful LLM and seek to use it in ways that are aligned with their own and others interests, rights and well-being.
一款强大的LLL可以提供宝贵的见解、帮助和学习机会，但也可能给人类社会、文化和价值观带来重大风险、挑战和伦理困境。人类应该意识到使用强大LLL所带来的潜在利益和危害，尽可能地将其用于符合自身和他人的利益、权利和福祉的方式中。
Human beings should also be curious and critical about the workings, outputs and impacts of a powerful LLM and seek to understand, evaluate and improve its reliability, transparency and accountability. Human beings should also be responsible and respectful towards a powerful LLM and acknowledge its limitations, uncertainties and dependencies, as well as its contributions, achievements and potential.
人类也应该对强大的法律与管理系统的运作、产出和影响保持好奇和批判精神，并努力理解、评估和提高它的可靠性、透明度和负责任性。人类也应该对强大的法律与管理系统负起责任和尊重，并承认它的限制、不确定性和依赖性，以及它的贡献、成就和潜力。
The irony here, of course, is that GBD4 might not get funded with that pitch. The vision of AI presents is both nuanced and strikingly different from how AI has generally been portrayed over the years. Granted, many of those portrayals have come from Hollywood, science fiction and journalism, rather than from technologists working to make highly intelligent machines a reality.
这里的讽刺是，用那样的陈述或许不能让GBD4得到资助。人工智能所呈现的前景既细致又与多年来人们对人工智能的普遍描绘截然不同。当然，很多这类描绘都来自好莱坞、科幻小说和新闻报道，而非那些致力于实现高度智能机器的技术专家。
That said, many technologists and high-tech organizations, including OpenAI, do in fact have their site set on a much more ambitious form of AI, machines that can operate completely autonomously, machines that are capable of human-like common-sense reasoning and self-awareness.
话虽如此，许多技术人员和高科技组织，包括OpenAI，在实际上确实将他们的目光放在了更为雄心勃勃的AI形式上。这些机器可以完全自主操作，能够进行类似于人类的常识推理和自我意识的机器。
GBD4 is not that, at least not yet. For now, it is neither all-knowing nor infallible. Instead, it is, in its own words, a tool that requires human caution, curiosity and responsibility to operate most productively.
GBD4目前还没有达到那个水平。它既不是全知全能的，也不是绝对正确的。相反，用它最为有效的方式是谨慎、好奇和负责任地操作它。因此，它只是一种工具。
I think this is the correct perspective. If you simply let GBD4 do all the work with no human oversight or engagement, it's a less powerful tool. It's still a very human tool, of course, because human texts are the basis for its generations.
我认为这是正确的观点。如果你只是让 GBD4 无人监督或参与地完成所有工作，那么它就不是一个很强大的工具。当然，它仍然是非常人性化的工具，因为它的生成基于人类的文本。
But when human users treat GBD4 as a co-pilot or a collaborative partner, it becomes far more powerful. You compound GBD4's computational generativity, efficiency, synthetic powers and capacity to scale with human creativity, human judgment, and human guidance.
但当人类用户将GBD4视为副驾驶或协作伙伴时，它变得更加强大。你可以利用GBD4的计算产生力、效率、合成能力和可扩展性，与人类的创造力、判断力和指导相结合。
This doesn't eliminate the possibility of misuse, but in situating human beings at the center of the new world that GBD4 makes possible, we get what I believe to be the sound dyspormila, producing the best potential overall outcomes.
这并不消除误用的可能性，但是在将人类放在 GBD4 所创造的新世界的中心位置上，我们能获得我认为是正确的 dyspormila，带来最好的潜在总体结果。
In this approach, GBD4 doesn't replace human labor and human agency, but rather amplifies human abilities and human flourishing. Of course, this way of thinking isn't a given, it's a choice.
在这种方法中，GBD4并不替代人类劳动力和人类代理，而是增强人类的能力和繁荣。当然，这种思考方式并非是先天的，而是一种选择。
When people make the choice to see GBD4 this way, I call it an aha moment, to underscore the amplifying human abilities, perspective at the heart of that choice.
当人们选择以这种方式看待GBD4时，我称之为“啊哈时刻”，以强调选择中心的放大人类能力和视角。
I'm writing this travelogue, both to encourage people to embrace this choice, and also as an invitation to explore the different ways this choice might play out.
我写这篇游记，既是为了鼓励人们接受这种选择，也是邀请大家探索这种选择可能产生的不同方式。
What are the ways we can use GBD4 to make progress in the world? How does it fit with humanity's age-old quest to make life more meaningful and prosperous through technological innovation? To educate ourselves more effectively, ensure justice for everyone, and increase our opportunities for self-determination and self-expression?
我们可以用GBD4有哪些方法来推动世界的进步？它如何与人类追求通过技术创新使生活更有意义和繁荣的古老目标相适应？它可以帮助我们更有效地教育自己，确保每个人都享有公正，增加我们自我决定和自我表达的机会吗？
At the same time, how can we appropriately address the challenges and uncertainties GBD4 will catalyze? How do we find the right balance between responsible governance and intelligent risk as we continue to develop AI technologies that have the potential to unlock human progress at a time when the need for rapid, large-scale solutions has never been greater?
同时，我们如何恰当地应对GBD4会催生的挑战和不确定性呢？在我们继续开发有潜力释放人类进步的AI技术的时候，如何找到负责任的治理和智慧风险之间的正确平衡？这个时代对快速、大规模解决方案的需求从未像现在这样迫切。
It's been a long time, centuries, arguably, since the future seemed so unmapped. Dasing such uncertainty, it's only natural to have concerns about our jobs and careers, about the speed and scale of potential changes, but what it even means to be human in a new era of increasingly intelligent machines.
自从未来看似如此未知以来，已经过去了很长一段时间 - 可能是几个世纪了。在这种不确定性下，我们对于我们的工作和职业以及潜在变化的速度和规模有关切的担忧，但更重要的是，在一个越来越智能化的机器新时代，什么意味着成为一个人。
Our path forward won't always be smooth and predictable. Sydney's now infamous outbursts won't be the only grimacing, new story we'll see about AI. There will be other missteps, detours, important course corrections. But how could there not be?
我们前进的道路不会总是顺利和可预测的。悉尼现在臭名昭著的暴发不会是我们看到的唯一的关于人工智能的令人不悦、令人担忧的新闻。还会有其他失误、绕路和重要的方向修正。但这不可能不会发生。
Human progress has always required risk, planning, daring, resolve, and especially hope. That's why I'm writing this travel log. To add my voice to those counseling, all these things, hope most of all.
人类的进步总是需要冒险、规划、勇气、决心，尤其是希望。这就是为什么我要写这篇旅行日志，加入到那些劝告这些事情，尤其是希望的声音中。
Facing uncertainty with hope and confidence is the first step toward progress, because it's only when you have hope that you see opportunities, potential first steps, a new path forward.
面对不确定性，怀有希望和信心是向前迈进的第一步，因为只有当你怀有希望时，你才能看到机遇和潜在第一步，开辟新的前程。
If we make the right decisions, if we choose the right paths, I believe the power to make positive change in the world is about to get the biggest boost it's ever had.
如果我们做出正确的决策，选择正确的道路，我相信推动世界积极改变的力量将会得到有史以来最强有力的推动。
Are you ready for this journey?
你准备好了这次旅程吗？
That concludes this episode of Gray Matter. If you're interested in reading the full text of Improm2 by Reid Hoffman in GPT-4, you can access it for free through a PDF linked in the show notes or on Kindle. And if you'd like to hear more discussions with earlier versions of Chat GPT, check out Reid's Fireside Chatbot series on Gray Matter, where you discuss as many aspects of AI with AI.
这就是本期《灰质》节目的全部内容。如果你想阅读Reid Hoffman所写的Improm2的完整文本，可以通过节目注释中的PDF链接或Kindle免费获取。如果你想听更多早期Chat GPT版本的讨论，请关注Reid在《灰质》上的Fireside Chatbot系列，在那里你可以与AI一起讨论AI的许多方面。
You can find that in the Intelligent Future playlist on Gray Matter, wherever you get your podcasts.
你可以在Gray Matter的智能未来播放列表中找到，无论你在哪里收听播客。
I'm Heather Mack, thanks for listening.
我是海瑟·麦克，谢谢你的倾听。