These are synthetic IDs, as good as a genuine ID.

這些都是合成身份證，與真身份證一樣好用。

Didn't understand the phrase.

沒聽懂這句話。

He took it literally.

他從字面上理解了這一點。

Oh, it even read my handwriting.

哦，它甚至能看懂我的筆跡。

Is this what people mean when they say that AI is going to take our jobs?

人們說人工智能會搶走我們的工作，是不是這個意思？

But what does a day in one of these roles actually look like?

但是，這些角色的一天究竟是什麼樣的呢？

From fighting scams and fraud, developing large-language models from scratch, to designing chatbots, they're all working with one technology, AI.

從打擊詐騙和欺詐、從零開始開發大型語言模型到設計哈拉機器人，他們都在使用一種技術--人工智能。

This is Most Wanted.

這是最高通緝令。

I'm going to show you some cool stuff, Nisa.

我要給你看一些很酷的東西，妮莎。

Yeah, that's absolutely cool.

是啊，這絕對酷斃了。

That's a mask of your face.

這是你臉上的面具。

Yeah, it's an exact replica of how I look like.

是啊，這和我的樣子一模一樣。

Rajat Maheshwari is part of the Cyber and Intelligence Solutions team at Mastercard.

Rajat Maheshwari 是萬事達卡網絡與智能解決方案團隊的一員。

In his role, his team develops tools to manage customer risk and prevent scams.

他的團隊負責開發管理客戶風險和防止詐騙的工具。

Part of his role involves getting in the mind of potential fraudsters by creating fake identities.

他的部分工作是通過製造假身份進入潛在欺詐者的腦海。

So these are synthetic IDs, as good as a genuine ID.

是以，這些都是合成身份證，與真身份證一樣好用。

You can have these identities, you can have the face mask, you can have the fingerprint, and then you're essentially replicating someone else.

你可以擁有這些身份，你可以擁有面具，你可以擁有指紋，然後你基本上就是在複製別人。

That person can do anything with these.

那個人可以用這些東西做任何事情。

This was done by a Japanese artist who did the mask for the James Cameron movie Avatar.

這個面具是由一位日本藝術家製作的，他曾為詹姆斯-卡梅隆的電影《阿凡達》製作過面具。

Sometimes we have to think like bad actors to come up with the solution which can stop these things.

有時，我們必須像壞人一樣思考，才能想出阻止這些事情發生的辦法。

We have seen that the world has evolved from these masks and now the deepfakes are coming in.

我們已經看到，這個世界已經從這些面具中發展了出來，而現在，"深度偽造 "正在襲來。

The intent was not to break the technology, but the intent was to help the solution providers to enhance the level so that they can stand against these kinds of attacks as well.

這樣做的目的不是要破壞技術，而是要幫助解決方案提供商提高水準，使他們也能抵禦這類攻擊。

Wow, what do we have here?

哇，這是什麼？

Yes, please, please, please have a seat.

是的，請坐，請坐，請坐。

Over at Amazon Web Services, Joel Garcia and his team have made a game out of simulating real-world security conditions as well, which they hope will help clients improve their incident response processes.

在亞馬遜網絡服務公司（Amazon Web Services），喬爾-加西亞（Joel Garcia）和他的團隊也製作了一款模擬真實世界安全狀況的遊戲，希望能幫助客戶改進事件響應流程。

What we have here is a project called Chaos Kitty.

我們現在看到的是一個名為 "Chaos Kitty "的項目。

Why kitty?

為什麼是小貓咪？

Well, I have cats at home who always destroy my furniture.

我家裡養了貓，它們總是破壞我的傢俱。

Oh yes, agents of chaos, for sure.

哦，是的，肯定是混亂的代理人。

What exactly is chaos engineering?

混沌工程究竟是什麼？

We're going to intentionally inject some failure so that we can learn.

我們要有意識地注入一些失敗，這樣才能學到東西。

We can see that there's many colourful lights and all these bricks that we use to represent what we have in our AWS cloud.

我們可以看到，這裡有許多五顏六色的燈和所有這些磚塊，我們用它們來表示 AWS 雲中的內容。

When it's red, something is wrong with the security configuration and when it's green, it's all good, all compliant.

紅色表示安全配置出了問題，綠色表示一切正常，符合要求。

We added on a Gen AI assistant here.

我們在這裡添加了 Gen AI 助手。

What we have here is a typical company security policy.

這就是典型的公司安全政策。

Typically, without a Gen AI assistant, they would have to look through and study this policy document very deeply.

通常情況下，如果沒有 Gen AI 助手，他們就必須深入翻閱和研究這份政策文件。

We have fed this document along with the Gen AI assistant so it knows all this information.

我們已將這份文件輸入 Gen AI 助手，是以它知道所有這些資訊。

I could go in there and ask questions.

我可以進去問問題。

It's going to give me some best practices.

它會給我提供一些最佳實踐。

Alexa, fix Chaos Kitty.

Alexa，修好混沌貓。

Remediating Chaos Kitty environment.

修復混沌小鷹環境。

You'd be able to leverage AI and Gen AI to actually help fix the challenges and then they'd be more focused on the other areas that could be improved.

你可以利用人工智能和 Gen AI 來真正幫助解決挑戰，然後他們就會更專注於其他可以改進的領域。

To be able to converse naturally with an AI chatbot, a large-language model, or LLM, is needed.

要想與人工智能哈拉機器人自然對話，就需要一個大型語言模型，即 LLM。

LLMs are AI models, pre-trained on vast amounts of data, which can understand and generate human language responses.

LLM 是在大量數據基礎上預先訓練的人工智能模型，能夠理解並生成人類語言的反應。

Popular models include ChatGPT-4 and Gemini.

流行的型號包括 ChatGPT-4 和 Gemini。

The generative AI market is expected to grow over a trillion dollars in the next decade.

未來十年，生成式人工智能市場預計將增長超過一萬億美元。

The team in Singapore is developing a large-language model that's catering to Southeast Asian languages.

新加坡的團隊正在開發一種大語言模式，以滿足東南亞語言的需求。

Leong Wei Tee speaks 14 languages fluently and he's an AI engineer and linguist involved in developing the Sea Lion model.

Leong Wei Tee 精通 14 種語言，他是一名人工智能工程師和語言學家，參與了海獅模型的開發。

We're going to look at an example about informal Indonesian.

我們來看一個關於非正式印尼語的例子。

We're essentially asking the model, our friend is sort of, because of his work, he's sort of panicking, and he's working very hard every day.

我們基本上是在問這位模特，我們的朋友因為工作的關係，有點慌亂，每天都很努力地工作。

How can we help him best manage his work?

我們如何幫助他更好地管理自己的工作？

Yeah, but in that context, they're saying, they're using an idiom, right?

是的，但在這種情況下，他們在說，他們在用一個成語，對嗎？

And they're saying that his beard is always on fire. So that idiom might not be so understandable to certain models, right?

他們還說他的鬍子總是著火。 所以這個成語對某些模特來說可能不是那麼好理解，對嗎？

So let's see how they deal with this.

讓我們看看他們是如何處理這個問題的。

For this demo, we have four panels, and each of them corresponds to one language model.

在這個演示中，我們有四個面板，每個面板對應一個語言模型。

So on the left, we have Sea Lion, our model, and we have three other models on the right.

左邊是我們的模型 "海獅"，右邊是其他三個模型。

We can see that it didn't understand the phrase.

我們可以看出，它並不理解這句話。

It took it literally, basically.

它基本上是照字面意思理解的。

Like it's thinking about setting someone on fire.

就像它想放火燒人一樣。

Oh no, it's also in the same tone. It's casual, informal, it's still colloquial, whereas this one is still sticking to like a more formal kind of response.

哦，不，語氣也一樣。 它是隨意的、非正式的，仍然是口語化的，而這個還在堅持一種更正式的迴應。

And that's really what we want to achieve as well.

這也正是我們想要實現的目標。

Now we are seeing that some of these models out there, they are not able to sort of handle multicultural contexts.

現在，我們看到，現有的一些模式無法處理多元文化背景。

And that's sort of understandable because they're building those models for a particular audience, right?

這也是可以理解的，因為他們是在為特定的閱聽人打造這些模型，對嗎？

For us in Southeast Asia, we need to operate within this region, handling our languages and cultures.

對於我們東南亞國家來說，我們需要在這個地區內開展業務，處理我們的語言和文化。

So this is why we decided to build Sea Lion.

這就是我們決定建造海獅號的原因。

Vincent Oh works as a Senior Specialist Solutions Architect at Amazon Web Services, and his projects involve leveraging generative AI based on human prompts to create personalized experiences.

Vincent Oh 在亞馬遜網絡服務公司擔任高級專家解決方案架構師，他的項目涉及利用基於人類提示的生成式人工智能來創建個性化體驗。

StoryGen was a project that we did with the National Library Board of the entire Singapore, where we wanted to reinvent the future of the libraries.

StoryGen 是我們與整個新加坡的國家圖書館委員會合作開展的一個項目，我們希望重塑圖書館的未來。

We used generative AI and AWS technology to create an experience whereby young children and adults, they can put in a series of inputs, a selection, and a brand new book will be created on the spot for them.

我們利用生成式人工智能和 AWS 技術創造了一種體驗，讓幼兒和成人可以輸入一系列資訊，進行選擇，然後當場為他們創建一本全新的書。

What you call the prompts, right, that you're actually sending back to the large language model.

你所謂的提示，實際上就是你發送回大語言模型的內容。

This is just the very beginning of the art of the possible, and it's going to be amazing how people will leverage gen-AI to unleash their extended level of creativity.

這僅僅是 "可能的藝術 "的開端，人們將如何利用人工智能技術釋放出更大的創造力將令人驚歎。

Part of those new jobs that are created as part of AI is a role or a skill set which is actually prompt engineering, which didn't exist before.

人工智能創造的新工作崗位中，有一部分是一種角色或技能組合，這實際上是一種提示工程，以前並不存在。

And when you look at prompt engineering, you don't need to be technical, you just need to understand how to put those prompts as you would, for example, like a detailed search, to maximize and leverage the power of a large language model.

當你在研究提示工程時，你不需要技術，你只需要瞭解如何將這些提示放置得像詳細搜索一樣，以最大限度地利用大型語言模型的力量。

There is a growing demand for AI specialists, and here at AI Singapore, the rest of Wei Qi's team, hailing from all over Southeast Asia, including Vietnam, the Philippines and Thailand, are working on making Sea Lion more inclusive.

對人工智能專家的需求日益增長，在新加坡人工智能公司，Wei Qi 團隊的其他成員來自東南亞各地，包括越南、菲律賓和泰國，他們正在努力使海獅公司更具包容性。

You are the one who brought everyone into this team.

是你讓大家加入這個團隊的。

More or less, more or less.

或多或少，或多或少。

My dream is eventually we have everybody, you know, from all the countries in Southeast Asia being represented here.

我的夢想是，我們最終能讓東南亞所有國家的所有人都來這裡參加會議。

AI Singapore's apprenticeship program aims to grow the pool locally, even those considering mid-career switches.

AI 新加坡的學徒計劃旨在擴大在地人才庫，甚至包括那些考慮中途轉行的人。

In fact, some of the Sea Lion team graduated as AI apprentices after switching disciplines, such as Tai, who studied finance, and Wei Qi, who was previously a pharmacist.

事實上，海獅團隊中的一些人是在轉專業後作為人工智能學徒畢業的，例如學金融的戴同學和曾是藥劑師的魏琦。

We're not really alone in trying to tackle low-resource languages.

在解決低資源語言問題上，我們並不孤單。

Ultimately, all the stuff that we do is fully open-source as well, and it's really shared with the public in general so that everyone can benefit.

最終，我們所做的一切也都是完全開源的，真正做到了與公眾共享，讓每個人都能從中受益。

Everyone is working on different aspects of the large language model.

每個人都在研究大語言模型的不同方面。

We get to learn about it with each other.

我們可以互相學習。

It does feel a bit like mini ASEAN.

這確實有點像小型東盟。

Mini ASEAN, yeah.

迷你東盟，是的。

There is so much to do in this field.

在這一領域有太多的事情要做。

Yeah, and AI, you know, is one of those industries where you can make a mid-career switch.

是的，你知道，人工智能是一個可以在職業生涯中期轉換的行業。

Absolutely, absolutely.

當然，當然。

And I'm a good example, Nisa.

我就是個好榜樣，妮莎。

Rajat started his career in the mobile and semiconductor industry and made the switch over to AI in 2014.

Rajat 的職業生涯始於移動和半導體行業，並於 2014 年轉向人工智能領域。

I have seen that journey in the past 20 years or so where AI has really flourished and it's becoming an inherent part of our lives.

在過去的 20 多年裡，我看到了人工智能真正蓬勃發展的歷程，它正在成為我們生活中固有的一部分。

When you ask people to use the digital ecosystem, there are side effects.

當你要求人們使用數字生態系統時，就會產生副作用。

According to the Global Anti-Scam Alliance, more than $1 trillion are lost to scams every year, affecting 2 billion victims.

根據全球反詐騙聯盟（Global Anti-Scam Alliance）的數據，每年因詐騙而損失的資金超過 1 萬億美元，受影響的受害者達 20 億人。

Rajat's work at Mastercard includes working with banks and governments to apply intelligent AI systems to predict whether scams are taking place.

Rajat 在萬事達卡的工作包括與銀行和政府合作，應用智能人工智能系統預測是否發生詐騙。

Let's take the example that you giving me $100, Nisa, looks a perfectly legitimate transaction.

舉個例子，你給我 100 美元，Nisa，看起來是一筆完全合法的交易。

But 1,000 more people giving me $100 in a day, something is suspicious.

但是，一天之內有 1000 多人給我 100 美元，這就有些可疑了。

There were sound alarm bells.

警鐘長鳴。

Exactly.

沒錯。

We train the models to detect the behavior, to detect the patterns, and then do the risk assessment and save you being sending money to me.

我們對模型進行訓練，以檢測行為、檢測模式，然後進行風險評估，讓你不用再把錢寄給我。

Welcome to the Mastercard Experience Center.

歡迎訪問萬事達卡體驗中心。

Whenever a person taps the card, our AI models kicks in, then we give a risk assessment to the issuing institution that what do we feel about this transaction.

只要有人刷卡，我們的人工智能模型就會啟動，然後向髮夾機構提供風險評估，說明我們對這筆交易的看法。

You can see the screen here showing that the transactions are getting declined.

您可以看到屏幕顯示交易被拒絕。

Where do you see the state of AI in the next couple of years?

您認為未來幾年人工智能的發展前景如何？

The integration with the LLMs.

與 LLMs 的整合。

So that's, LLM is essentially the large language models.

是以，LLM 本質上就是大型語言模型。

Large language models form a big part of the user experience solutions that Joel works on for AWS.

大型語言模型是喬爾為 AWS 研發的用戶體驗解決方案的重要組成部分。

Take reading invoices, for example.

以閱讀發票為例。

If you want to introduce something else into your business, a new form, a new process, then you'll have to reevaluate the technology or refactor that in.

如果你想在業務中引入其他東西、新的形式、新的流程，那麼你就必須重新評估技術或重構技術。

With Gen AI, because of the large language model, it can actually parse these.

有了 Gen AI，因為有了大型語言模型，它就能真正解析這些內容。

Parsing refers to the process of breaking down a user's input into smaller pieces and analyzing each piece to determine its meaning.

解析是指將用戶輸入的內容分解成更小的片段，並分析每個片段以確定其含義的過程。

I'd like to know if there is a proper signatory.

我想知道是否有適當的簽字人。

Oh, it even read my handwriting, so it knows that it's my terrible handwriting, that it's Michael Garcia.

哦，它甚至能看懂我的筆跡，所以它知道這是我糟糕的筆跡，是邁克爾-加西亞的筆跡。

Did you see yourself ending up where you are today?

你預見到自己今天的結局了嗎？

Even at a young age, I knew that in some form, way, or that I was going to be involved with technology.

即使在很小的時候，我就知道自己會以某種形式、某種方式與技術打交道。

What did you start out studying?

你最初學的是什麼？

I graduated computer science.

我畢業於計算機科學專業。

I went through a variety of jobs.

我做過各種各樣的工作。

I even founded a company before startups were fashionable.

我甚至在創業公司成為時尚之前就創辦了一家公司。

Today, with AI, when you're thinking about redefining experiences, you're not going to get to it on the first.

如今，有了人工智能，當你在考慮重新定義體驗時，就不會在第一時間去做了。

This is like super long-shot ambition, right?

這是個超級遠大的理想，對嗎？

You know that Southeast Asia is not 10 languages, only there are like hundreds of dialects, Balinese, Javanese, Visayan.

要知道，東南亞並沒有 10 種語言，只有數百種方言，如巴釐語、爪哇語、米沙鄢語。

So, we hope to get those represented as well, eventually.

是以，我們希望最終也能有這些代表。

How soon can we get there?

我們多快能到達那裡？

Our hope is that it will trigger similar movements across the region.

我們希望，這將在整個地區引發類似的運動。

Outside of Southeast Asia, the interest is also very strong, right?

在東南亞之外，人們的興趣也非常濃厚，對嗎？

We have similar situation in India, we have similar situation in Africa, where there's a bit of under-representation.

印度也有類似的情況，非洲也有類似的情況，都存在代表性不足的問題。

If I'm able to save someone's lifelong savings, I mean, it's itself that makes my day.

如果我能挽救一個人的畢生積蓄，我的意思是，這本身就是我的一天。

Is this what people mean when they say that AI is going to take our jobs?

人們說人工智能會搶走我們的工作，是不是這個意思？

It's going to make something more efficient, so that people can actually concentrate on what matters more, either for the business or for customers.

它將提高工作效率，使人們能夠真正專注於更重要的事情，無論是對企業還是對客戶。