全文解析：OpenAI GPT-4 发布会内容

😄看看Greg Brockman,President and Co-Founder of OpenAI全程演示都说了什么~

Honestly, it's kind of hard for me to believe that this day is here. OpenAI has been building this technology really since we started the company, but for the past two years, we've been really focused on delivering GPT-4. That started with rebuilding our entire training stack（意思是重新构建整个机器学习模型训练的技术和基础设施。这个过程可能涉及升级硬件、软件和流程，以提高训练过程的速度、效率和准确性）, actually training the model, and then seeing what it was capable of, trying to figure out its capabilities（能力，"Ability" 更侧重于个体的潜力和独特性，而 "Capability" 更强调一种更广泛的能力、知识和技术）, its risks, working with partners in order to test it in real-world scenarios, really tuning its behavior, optimizing the model, getting it available so that you can use it.

And so today, our goal is to show you a little bit of how to make GPT-4 shine, how to really get the most out of it（尽可能充分利用）, you know, where its weaknesses are, where we're still working on it, and just how to really use it as a good tool, a good partner. 

第一个案例：文本总结与合成

So the first thing I want to show you is the first task that GPT-4 could do that we never really got 3.5 to do. And the way to think about this is all throughout training, you know, you're constantly doing all this work. It's 2 a.m, the pager goes off, you fix the model, and you're always wondering, is it gonna work? Is all this effort actually going to pan out（work out or succeed）? And so we all had a pet task（"a pet task" 是一个英语习语，表示一个人喜欢做的、个人特别关注的任务或项目，通常是因为它与个人的兴趣、技能或价值观相关。这个词组中的 "pet" 一词表示 "宠物"，暗示了这个任务或项目是像宠物一样被人们照顾和关注的。这个词组通常用于形容一个人在工作或学习中的任务或项目，也可以用于形容一个人在日常生活中的爱好或兴趣。） that we really liked and that we would all individually be trying to see, is the model capable of it now?

The way that it works is you have a system message where you explain to the model what it's supposed to do, and we've made these models very steerable（可以被控制或操纵的，eg：This video game controller features highly responsive and steerable joysticks for precise gameplay.）, so you can provide it with really any instruction you want, whatever you dream up, and the model will adhere to it（坚持、遵守某个规定、计划或决定） pretty well. And in the future, it will get increasingly powerful at steering the model very reliably.（未来，模型的控制系统将变得越来越强大，并且能够非常可靠地操纵模型） 

All right. So now, time to actually show you the task that I'm referring to. So everyone's familiar with "summarize this article into a sentence," okay, getting a little more specific, but where every word begins with G. So this is 3.5, let's see what it does.

Yeah, it kind of didn't even try, just gave up on the task. This is pretty typical for 3.5 trying to do this particular kind of task. If it's, you know, sort of a very kind of stilted article（写作语言和风格过于正式、生硬或不自然的文章） or something like that, maybe it can succeed, but for the most part, 3.5 just gives up. But let's try the exact same prompt, the exact same system message in GPT-4. 

So, kind of borderline（含糊不清的） whether you want to count AI or not, but let's say AI doesn't count. That's cheating, so fair enough. The model happily accepts my feedback. 

（接下来还试了A，Q开头，过程略）

All right, so I've shown you summarizing an existing article. I want to show you how you can flexibly combine ideas between different articles. So, I'm going to take this article that was on Hacker News yesterday, copy-paste it into the same conversation so it has all the context of what we're just doing. I'm going to say, find one common theme between this article and the GPD4 blog.

So, this is an article about Pinecone, which is a Python web app development framework, and it's making the technology more accessible, user-friendly（易于使用，用户友好）. If you don't think that was insightful（有深刻见解的） enough, you can always give some feedback and say that was not insightful enough. Please, no, I'll just even leave it there, leave it up to the model to decide.

So, bridging the gap between powerful technology and practical applications seems not bad. And of course, you can ask for any other kind of task you want. Using its flexible language understanding and synthesis（语言理解和合成能力）, you can ask for something like, now turn the GPD4 blog post into a rhyming poem（韵律诗）.

So, there we go. This is consuming existing content using GPT4 with a little bit of creativity on top. （加入点创造力）

第二个案例：代码生成，写了一个能接受图片和文字信息输入的Discord bot

But next, I want to show you how to build with GPT4. What it's like to create with it as a partner. And so, the thing we're going to do is we're going to actually build a Discord bot.

I'll build it live and show you the process, show you debugging（调试）, show you what the model can do, where its limitations are, and how to work with them in order to sort of achieve new heights（达到新高度：reach new levels of success, excellence, or accomplishment that surpass what has been previously achieved）. 

So, the first thing I'll do is tell the model that this time it's supposed to be an AI programming assistant. Its job is to write things out in pseudocode（伪代码） first and then actually write the code.And this approach is very helpful so that the model breaks down the problem into smaller pieces.（分解问题）

pseudo-是假的伪的，eg：pseudonym，笔名，she's written adult fiction under a pseudonym.

pseudocode指的是一种高级抽象层次的描述性程序设计语言，它不是任何特定编程语言的实现，而是一种与计算机无关的通用描述方法。它是一种类似于自然语言的表示方法，可用于描述程序的算法或逻辑流程。使用pseudocode的好处是可以让程序员更容易地思考和设计算法，并且可以跨越不同的编程语言和计算机体系结构进行实现。同时，pseudocode也可以作为编写代码前的预备工作，帮助程序员更好地规划代码的架构和结构。

And then, that way, you're not kind of asking it to just come up with a super hard solution to a problem all in one go. It also makes it very interpretable（可判断的） because you can see exactly what the model was thinking, and you can even provide corrections if you'd like.

So, here is the prompt that we're going to ask it. This is the kind of thing that 3.5 would totally choke on（卡住，难以应对） if you've tried anything like it. But, so, we're going to ask for a Discord bot that uses the GPT-4 API to read images and texts.

Now, there's one problem here, which is this model's training cutoff（通常指机器学习模型所使用的训练数据的截止日期） is in 2021, which means it has not seen our new chat completions format. So, I literally just went to the blog post from two weeks ago, copy-pasted from the blog post, including the response format. It has not seen the new image extension to that, and so I just kind of wrote that up, and you know, just very minimal detail about how to include images. So, now, the model can actually leverage the documentation that it did not have memorized that it does not know.（提供了一些新的信息，使模型能够更好地理解如何在文本中包含图片）

And, in general, these models are very good at using information that it's been trained on in new ways and synthesizing new content. And, you can see that right here, that it actually wrote an entirely new bot.

Now, let's actually see if this bot is going to work in practice. So, you should always look through the code to get a sense of what it does. Don't run untrusted code from humans or from AIs.

And, one thing to note is that the Discord API has changed a lot over time, and particularly that there's one feature that has changed a lot since this model was trained.

Give it a try. In fact, yes, we are missing the "intents" keyword. This is something that came out in 2020.

So the model does know it exists, but it doesn't know which version of the Discord API we're using. So, are we out of luck? Well, not quite. We can just simply paste the error message to the model. We're not even going to say, "Hey, this is from running your code. Could you please fix it?" We'll just let it run.

And the model says, "Oh yeah, whoops! The 'intense' argument here's the correct code." Now, let's give this a try again, making sure that we understand what the code is doing.

//这里遇到了第二个问题，AI 不知道代码运行的环境

Now, a second issue that can come up is that it doesn't know what environment I'm running in. If you notice, it says, "Hey, here's this inscrutable error message（难以理解的错误信息；inscrutable：impossible to understand or interpret）," which, if you've not used Jupyter Notebook a lot with AsyncIO before, you probably have no idea what this means. But fortunately, once again, you can just sort of say to the model, "Hey, I am using Jupiter and would like to make this work. Can you fix it?"（//是我这种菜鸟的福音了哈哈^_^）

So now we'll run, and it looks like something happened. 

第三个案例：对图像的描述

The first thing I'll do is go over to our Discord and I will paste in a screenshot of our Discord itself.

So, remember, GPT-4 is not just a language model, it's also a vision model. In fact, it can flexibly accept inputs that intersperse images and text arbitrarily, kind of like a document.（GPT-4 的新特性：它不仅仅是一个语言模型，还可以处理图像和文本混合的输入，就像处理文档一样）

Now, the image feature is in preview（预演阶段）, so this is going to be a little sneak peek（偷看，提前看）. It's not yet publicly available（公开）. It's something we're working with one partner called Be My Eyes in order to really start to develop it and get it ready for prime time. （面向广大用户使用，prime time：广播电视的黄金时段）

But you can ask anything you like. For example, I can say, "GPT-4, hello world, can you describe this image in painstaking detail（详细的描述）?" Alright, which, first of all, think of how you would do this yourself. There are a lot of different things you could latch onto（抓住）, a lot of different pieces of the system you could describe. 

So it's a screenshot of a Discord application interface. Pretty good, didn't even describe it, it knows that it's Discord. It's probably Discord written there somewhere where it just kind of knows this from prior experience. Server icon label GPT4 describes the interface in great detail, talks about all the people telling me that I'm supposed to do Q, a very, very kind audience, and describes much of the notification messages and the users that are in the channel. And so there you go, that's some pretty good understanding.

（后面又解了一个bug，过程略。。）

I think this augmenting tool makes you much more productive, but it's still important that you are in the driver's seat and are the manager and knows what's going on.

（工具虽强，人还是要做主导）

Squirrels do typically eat nuts. We don't expect them to use a camera or act like a human, so I think that's a pretty good explanation of why that image is funny. So, I'm going to show you one more example of what you can do with this model.

第四个案例：根据原型图生成网页

So, I have here a nice hand-drawn mock-up of a joke website, definitely worthy of being put up on my refrigerator. So, I'm just going to take out my phone, literally take a photo of this mock-up, and then I'm going to send it to our Discord.

All right, I'm going to send it to our Discord, and this is, of course, the rockiest part（一个过程中最具挑战性、最艰难的阶段）: making sure that we actually send it to the right channel, which, in fact, I think maybe I did not send it to the wrong channel. It's funny; it's always the non-AI parts of these demos that are the hardest part to do.  And here we go, technology is now solved, and now we wait.

So, the thing that's amazing in my mind is that what's going on here is we're talking to a neural network, and this neural network was trained to predict what comes next. It played this game of being shown a partial document and then predicted what comes next across an unimaginably large amount of content.  

（神经网络被训练用于预测下一个可能出现的内容。它通过对大量内容进行训练，模拟了在看到部分文档内容后预测接下来会出现什么的过程）

And from there, it learns all of these skills that you can apply in all these very flexible ways. And so, we can actually take now this output, so literally we just said to output the HTML from that picture, and here we go, actual working JavaScript filled in the jokes 

This was the original of our mock-up, and so there you go, going from hand-drawn beautiful art, if I do say so myself, to a working website. And this is all just potential, right? 

You can see lots of different applications. We ourselves are still figuring out new ways to use this, so we're going to work with our partner. We're going to scale up（扩大、增加或提高规模、范围或数量） from there, but please be patient because it's going to take us some time to really make this available for everyone.

one last thing：处理报税问题

So, I have one last thing to show you. I've shown you reading existing content, I've shown you how to build with the system as a partner. The last thing I'm going to show is how to work with the system to accomplish a task that none of us like to do but we all have to. So, you may have guessed the thing we're going to do is taxes.

Now, note that GPT is not a certified tax professional（美国报税异常复杂，所以会有专业的税务师）, nor am I, so you should always check with your Tax Advisor. But it can be helpful to understand some dense content（复杂的内容：material that is difficult to understand or requires a lot of effort to comprehend） to just be able to empower yourself to（赋能哈哈） solve problems and get a handle on （可以去处理to understand and be able to deal with ）what's happening when you could not otherwise.

So once again, I'll do a system message. In this case, I'm going to tell it that it's tax GPT, which is not a specific thing that we've trained into this model. You can be very creative if you want with the system message to really get the model in the mood of （在...情景下）what is your job, what are you supposed to do.

So, I pasted in the tax code. This is about 16 pages worth of tax code, and there's this question about Allison and Bob. They got married at one point, and here are their incomes, and they take a standard deduction（指的是可以从纳税人的总收入中减去的固定金额，以确定纳税人应该缴纳的可征税收入）. They're filing jointly. So first question: what is their standard deduction for 2018?

If you notice, the model got to the same conclusion, and you can actually read through its explanation. And to tell you the truth, the first time I tried to approach this problem myself, I could not figure it out. I spent half an hour reading through the tax code, trying to figure out this back reference and why there's some program like just what's even going on. It was only by asking the model to spell out its reasoning, and then I followed along, that I was like, oh, I get it now. 

I understand how this works. And so that I think is where the power of the system lies. It's not perfect, but neither are you, and together is this amplifying tool that lets you just reach new heights, and you can go further. 

（它并不完美，但你也不完美，而使用它就像一个放大器，让你可以达到新的高度，你可以走得更远）

And so to end it, the final thing that I will show is a little other dose of creativity, which is now summarize this problem into a rhyming poem. And there we go, a beautiful, beautiful poem about doing your taxes. 

So thank you everyone for tuning in！