Software Architecture for Gen AI by Rebecca Parsons

章节 1：GenAI 浪潮与架构师的新挑战

📝 本节摘要：

本节作为开篇，Rebecca Parsons 首先重新界定了演讲的主题范围，不仅关注软件架构，更探讨架构本身如何支持生成式 AI（GenAI）系统的部署，以及这对架构师角色的影响。她反驳了“架构师无用论”，强调该职位的正当性。尽管当前 AI 炒作（Hype）盛行，但她认为凭借摩尔定律带来的算力提升和海量数据的积累，我们拥有比以往更坚实的基础，不太可能重蹈“AI 寒冬”的覆辙。最后，她幽默地提醒听众，该领域变化极快，技术细节可能迅速过时，但底层逻辑依然适用。

[原文] [Rebecca Parsons]: I've changed the title slightly because as I was working on this talk it occurred to me that it's not just software architecture that is being impacted and will have an impact from gen AI but it's architecture more broadly I want to talk about not just architecture or software architecture but how architectures will support the deployment of gen systems and what are the implications on architecture from geni but also on architects now some people from time to time have said that I don't like Architects that I have something against Architects that is completely false I do believe that there are roles for people who call themselves architecture and just because some people perhaps don't do the job in the way that they should does not mean that the job is not legitimate and so I do want to talk about how this is going to also affect being a working architect now one of the things I want to start with is the fact that that there's so much hype and I've been involved with artificial intell

[译文] [Rebecca Parsons]: 我稍微修改了一下标题，因为我在准备这次演讲时意识到，不仅仅是软件架构正在受到生成式 AI（GenAI）的影响并将对其产生影响，而是更广泛意义上的架构。我不仅想谈论架构或软件架构，还想谈谈架构将如何支持生成式系统的部署，以及生成式 AI 对架构乃至架构师本人的影响。有些人时不时会说我不喜欢架构师，说我对架构师有成见，这完全是错误的。我确实相信那些自称为架构师的人是有其角色的，仅仅因为有些人可能没有以应有的方式完成工作，并不意味着这份工作本身是不正当的。因此，我确实想谈谈这将如何影响作为一名在职架构师的工作。我想从这一点开始：现在的炒作（hype）实在太多了，而我涉足人工智能领域……

[原文] [Rebecca Parsons]: igence um about since fire was invented it feels like sometimes and we've seen all kinds of AI Winters this doesn't feel the same and I think there's some reasons for that even though the hype is absolutely enormous what we have is a much stronger Foundation to work from part of this comes from Moore's Law computers are faster now isn't that wonderful memory is cheaper isn't that wonderful um but also just the enormous amount of data and we'll talk a little bit more about data but it does feel like even though there are going to people who disappointed because the hype is so great about what geni can do I do believe that we are not getting ready to go into an AI winter it isn't AI Autumn at the moment now of course with the pay pace of change much of what I'm going to tell you is going to be obsolete on Monday uh so you know take it with a grain of salt seriously though I'm going to try to talk about this in such a way that maybe the tools are going to be different maybe new algorithms

[译文] [Rebecca Parsons]: ……有时感觉就像自从火被发明以来我就在那儿了。我们要见过各种各样的“AI 寒冬”，但这次感觉不一样，我认为这是有原因的。尽管炒作绝对是铺天盖地的，但我们拥有一个更坚实的工作基础。这部分源于摩尔定律，计算机现在更快了，这不是很棒吗？内存更便宜了，这不是很棒吗？但也源于海量的数据，我们稍后会多谈一点关于数据的问题。确实感觉，尽管有些人会因为对生成式 AI 能力的炒作过高而感到失望，但我真的相信我们并没有准备进入一个“AI 寒冬”，此刻也不是“AI 之秋”。当然，随着变化的步伐，我今天要告诉你们的很多内容到了周一可能就过时了，所以你知道，对这些内容要持保留态度。说正经的，我将尝试以一种即使工具变得不同、即使发现了新算法……的方式来谈论这个话题。

[原文] [Rebecca Parsons]: are going to be discovered but a lot of the fundamentals if you're you're talking about a technology that is legitimately called genitor of AI it will still have some of these properties

[译文] [Rebecca Parsons]: ……但如果你谈论的是一项可以正当地被称为生成式 AI 的技术，很多基本原理仍然会具备这些属性。