跟讀練習: What is AI governance? - 通過影片學習英語口語
選擇練習模式
每種模式都會開啟全屏練習室。
34 句
如果句子過短或過長，請點選 Edit 進行調整。
1
Hello, my name is Kush Varshney.
2
I'm a distinguished research scientist and senior manager at IBM Research.
3
I'm here to answer three questions about AI governance.
4
Governance is an interesting word.
5
It means control.
6
And if we think back in history, steam engines had this part called the governor and it regulated the flow of steam
7
and it was there to make sure that the system remains safe.
8
And when we say AI governance, it's the set of practices to keep an AI system under control so that it remains safe as well.
9
So it includes things like respecting regulations, curating data, and creating fact sheets that explain how the system works.
10
Some of the risks posed by generative AI and LLMs are actually the same as traditional machine learning.
11
So things like fairness, transparency, robustness to attacks, they're all still concerns.
12
But then there's some new ones as well.
13
So new risks that are emerging with generative AI include things like toxicity, harmful behaviors, bullying, these sort of things.
14
And then there's also hallucination.
15
So where the model is putting out things which kind of sound like they would make sense, they're plausible, but they're actually factually incorrect.
16
And we've created a taxonomy of common harms that can mitigate these behaviors with data curation, with fine tuning, prompt tuning, prompt engineering.
17
So all of that is really the way to address them.
18
But this is just a starting point.
19
So if we start and think for a second,
20
governments, enterprises, all sorts of other organizations will have additional constraints to build into their LLMs so that they follow applicable laws,
21
industry standards, social norms, and so on.
22
And understanding the models is also quite different
23
because we no longer need to know exactly how the models make their decisions and predictions and predictions.
24
This is what we used to call explainability, but now instead what we need to be able to do
25
is trace an LLM's generative output to the user's prompt or to its training data.
26
So source attribution is the new explainability.
27
Safety guarantees with generative AI are very difficult because there's no one-size-fits-all approach.
28
Every organization needs to define their own values, their own behaviors that they find acceptable and unacceptable.
29
For example, once you know the context of your chatbot's deployment, you can start to reason through the relevant risks.
30
This could involve taking existing policy documents like laws,
31
corporate policies, or other rules and using them as instruction data to teach the model how to behave.
32
And interventions must also be scalable because we're dealing with such large models and even larger data sets.
33
And at IBM we're designing tools to address the unique threats posed by large language models and other content generating foundation models.
34
We're also devising new methods to limit chatbots from leaking personal or proprietary data and hallucinating wrong or irrelevant information.
同頻道
FOR YOU
推薦影片
C12
250 句
AI Strategy, Policy, and Governance | Allan Dafoe
Business EnglishUniversity StudentsWorking Adults
C111
98 句
What are AI Agents?
Business EnglishUniversity StudentsWorking Adults
C13
208 句
AI & Education: Generative AI & the Future of Critical Thinking
Business EnglishUniversity StudentsWorking Adults
C13
31 句
Deputies seek clarity on AI and IP rights | Vietnam Today
Daily ConversationUniversity StudentsWorking Adults
C143
6 句
Ray Dalio on the AI Bubble
Business EnglishUniversity StudentsWorking Adults
C11
83 句
Leading AI CEOs Warn UN Security Council of Existential Risks | Sam Altman On AI Controls | N18G
Business EnglishDaily ConversationUniversity Students
C11
402 句
FULL DISCUSSION: Google's Demis Hassabis, Anthropic's Dario Amodei Debate the World After AGI | AI1G
IELTS PrepBusiness EnglishUniversity Students
C13
219 句
Why AI is Legal’s New Competitive Advantage - Future of Work Podcast
Business EnglishUniversity StudentsWorking Adults
C11
127 句
The Alignment Problem Explained: Crash Course Futures of AI #4
Business EnglishDaily ConversationUniversity Students
![What is AI and ML? [Pt 2] | Generative AI with .NET for Beginners](https://img.youtube.com/vi/Len8KRBFvGo/mqdefault.jpg)
B22
52 句
What is AI and ML? [Pt 2] | Generative AI with .NET for Beginners
Business EnglishUniversity StudentsWorking Adults
C11
114 句
Harnessing generative AI in architecture, engineering & construction
Daily ConversationUniversity StudentsWorking Adults
真實場景：解讀AI治理
影片中，IBM研究人員以通俗易懂的方式解釋“AI治理”，從歷史上的蒸汽機調速器切入，對比傳統機器學習與生成式AI的風險，還提到企業如何通過資料管理、模型調優應對挑戰。這樣的科技對話場景在學術討論、職場彙報中很常見，既能學習專業表達，也能鍛鍊聽懂複雜概念的能力。
實用表達與搭配
- keep...under control：使……處於控制之中。例：AI governance aims to keep AI systems under control.（AI治理旨在讓AI系統處於可控狀態。）
- pose risks：帶來風險。例：Generative AI poses new risks like hallucination.（生成式AI帶來了幻覺等新風險。）
- trace...to：追溯……到。例：We need to trace the output to training data.（我們需要將輸出追溯到訓練資料。）
- one-size-fits-all approach：一刀切的方法。例：Safety guarantees need no one-size-fits-all approach.（安全保障不需要一刀切的方法。）
影子跟讀挑戰：提升口語與理解
影子跟讀（shadowspeak）是提高英語發音和流暢度的有效方法。現在就試試：播放影片中“Governance is an interesting word...so that it remains safe as well”這段，暫停後重復，注意模仿重音（如“control”“safe”）和語調。完成後，用“keep...under control”造句，描述一個生活場景（如“Parents try to keep their kids' screen time under control”）。通過這種“看影片學英語+即時練習”的方式，既能掌握專業詞彙，又能強化口語表達。記住，shadow speak的關鍵是“同步”與“模仿”，堅持一週，你會發現發音和聽力理解都有明顯進步。
什麼是跟讀法？
跟讀法 (Shadowing) 是一種有科學依據的語言學習技巧，最初開發用於專業口譯員的培訓，並由多語言者Alexander Arguelles博士普及。這個方法簡單而強大：您在聽英語母語原聲的同時立即大聲重複——就像是一個延遲1-2秒緊跟說話者的影子。與被動聽力或語法練習不同，跟讀法強迫您的大腦和口腔肌肉同時處理並模模擬實的講話模式。研究表明它能顯著提高發音準確性，語調，節奏，連讀，聽力理解和口語流利度——使其成為雅思口語備考和真實英語交流最有效的方法之一。