Interactive transcript · 146 lines
Hello, everyone.
大家好，各位好。
Welcome back to TalkCast, your favorite space for deep, friendly conversations about technology,
欢迎回到 TalkCast，这是您最喜欢的关于科技进行深刻、友好对话的平台，
science, and the ideas shaping our world.
科学以及正在塑造我们世界的思想。
I am Emily, your host.
我是艾米莉，您的主持人。
First, a huge, heartfelt thank you to our incredible TalkCast members.
首先，向我们不可思议的 TalkCast 会员致以巨大的衷心感谢。
Your support keeps this show alive and allows us to bring you these fascinating stories.
您的支持让这个节目维持下去，并让我们能够为您带来这些迷人的故事。
If you have not joined our membership program yet, please consider doing so.
如果您尚未加入我们的会员计划，请考虑加入。
TalkCast needs your support, and we have some wonderful benefits waiting for you.
TalkCast 需要您的支持，我们有一些丰厚的福利等待着您。
Today, we are diving into a major story that has sent shockwaves through the global scientific community.
今天，我们将深入探讨一个在全球科学界引发震波的重大新闻。
On September 11th, 25 Fields Medalists, who are often called the Nobel Prize winners of mathematics,
9月11日，25位菲尔兹奖得主（通常被称为数学界的诺贝尔奖获得者），
published a serious warning.
发表了一份严正警告。
They state that artificial intelligence companies and the mathematical community are facing a deep
他们声明，人工智能公司与数学界正面临着深刻的
and worrying misalignment.
和令人担忧的分歧。
Some are even asking, is AI ruining mathematics?
有些人甚至在问，AI 是否正在毁掉数学？
To help us understand what is going on, we are joined today by Dr.
为了帮助我们理解正在发生的事情，今天我们邀请到了
Charles, a professor of mathematics and an expert in AI ethics.
查尔斯博士，他是一位数学教授，也 是 AI 伦理领域的专家。
Welcome to the show, Charles.
欢迎来到节目，查尔斯。
Thank you, Emily.
感谢你，艾米莉。
It is a real pleasure to be here with you and our listeners today.
今天能和您以及我们的听众在一起，真是莫大的荣幸。
This is a very important conversation.
这是一场非常重要的对话。
It really is.
确实如此。
Charles, let us start with the basics.
查尔斯，让我们从基础谈起。
Who are these 25 people, and why does their warning carry so much weight?
这25个人是谁？为什么他们的警告具有如此大的分量？
Well, Emily, the Fields Medal is the highest honor a mathematician can receive.
嗯，艾米莉，菲尔兹奖是数学家所能获得的最高荣誉。
It is only awarded once every four years to mathematicians under the age of 40.
它每四年才颁发一次给40岁以下的数学家。
So the people who signed this warning are the absolute best minds in our field, including famous
因此，签署这份警告的人都是我们领域中最绝对顶尖的人才，包括著名的
figures like Terence Tao.
像陶哲轩这样的人物。
When 25 of the world's top mathematicians stand together to warn us about something, the entire
当25位世界顶尖数学家站在一起就某事向我们发出警告时，整个
scientific world stops and listens.
科学界都会停下来倾听。
They are the guardians of mathematical truth, and they are genuinely worried.
他们是数学真理的守护者，而且他们确实感到担忧。
That is incredibly powerful.
那真是极其有力量。
The warning uses the word misalignment.
这份警告使用了错位这个词。
In plain English, what does this mean?
用浅显的语言来说，这是什么意思？
How are AI companies and mathematicians misaligned?
人工智能公司和数学家是如何产生分歧的？
Simply put, Emily, misalignment means their goals and values do not match.
简单来说，艾米莉，错位意味着他们的目标和价值观不匹配。
AI companies want to build products quickly, make a profit, and move fast.
人工智能公司想要快速开发产品、获取利润并快速推进。
On the other hand, mathematicians value deep understanding, absolute truth, and open sharing.
另一方面，数学家重视深刻的理解、绝对的真理和公开的分享。
AI companies need high -quality data to train their models.
人工智能公司需要高质量的数据来训练它们的模型。
They have been scraping mathematical research papers, textbooks, and online forums without asking
它们一直在抓取数学研究论文、教科书和在线论坛，而未进行
for permission, without paying the creators, and without giving any credit.
请求许可，不向创作者支付费用，也不给予任何署名。
The math community feels that their hard work is being taken to create commercial tools, and that
数学界觉得他们的辛勤工作被用来开发商业工具，并且
these tools might eventually change or even hurt the way scientific research is done.
这些工具最终可能会改变甚至损害科学研究的方式。
Ah, so it's a mix of a copyright problem and a cultural clash.
啊，所以这是版权问题和文化冲突的结合。
In school, our math teachers always told us to show your work.
在学校里，我们的数学老师总是告诉我们要展示解题过程。
But with AI, it seems different.
但在 AI 面前，情况似乎不同。
It feels like AI just gives an answer without explaining how it got there.
感觉 AI 只是给出了答案，却没有解释是如何得出的。
Is that part of the issue?
这也是问题的一部分吗？
You hit the nail on the head, Emily.
你正好切中要害，艾米莉。
That is exactly right.
一点没错。
We often call AI a black box.
我们经常称 AI 为黑盒。
When an AI solves a complex problem, it does not explain its logical steps in a way that humans can understand.
当人工智能解决一个复杂的问题时，它不会以人类能够理解的方式解释其逻辑步骤。
In mathematics, the final answer is only a small part of the value.
在数学中，最终答案只是其价值的一小部分。
The most important part is the proof, which is the step -by -step logical journey.
最重要的部分是证明，这是一个按部就班的逻辑过程。
That journey teaches us why something is true and connects different areas of science.
那段历程教会我们为什么某些事物是正确的，并将科学的不同领域联系起来。
If we just accept answers from a black box, we lose our understanding, and science cannot grow.
如果我们只是接受来自黑箱的答案，我们就会失去理解，科学也无法增长。
I see.
我明白了。
So if we rely too much on AI, we might become passive users of technology rather than active thinkers.
所以，如果我们过度依赖人工智能，我们可能会成为技术的被动使用者，而不是积极的思考者。
But Charles, we have also heard about some amazing breakthroughs.
但查尔斯，我们也听到了一些令人惊叹的突破。
Didn't some AI models from Google actually solve really tough geometry problems recently?
谷歌的一些人工智能模型最近不是真的解决了一些非常棘手的几何问题吗？
Yes, they did.
是的，它们做到了。
We must be fair and look at both sides.
我们必须保持公正，客观看待这两个方面。
Systems like Alpha Geometry and Alpha Proof have done remarkable work.
像 Alpha Geometry 和 Alpha Proof 这样的系统做出了卓越的工作。
They solved complex problems from the International Mathematical Olympiad.
它们解决了来自国际数学奥林匹克竞赛的复杂问题。
However, they did this using what we call formal languages.
然而，它们是通过使用我们所说的形式语言做到这一点的。
This means the AI had to translate the math into a special code that can be verified by a computer.
这意味着人工智能必须将数学转化为可以被计算机验证的特殊代码。
This process is called formal verification.
这个过程被称为形式验证。
Think of it like a super smart digital referee that checks every single line of code to make sure
可以把它想象成一个超级聪明的数字裁判，负责检查每一行代码以确保
there are no mistakes.
没有任何错误。
This is a wonderful tool for mathematicians because it helps us find errors quickly.
对于数学家来说，这是一个极好的工具，因为它能帮助我们快速发现错误。
But again, it is a tool.
但话又说回来，它只是一个工具。
It does not replace the human intuition needed to create new mathematical concepts in the first place.
它并不能取代最初创造新数学概念所需的人类直觉。
That is a very useful distinction.
这是一个非常实用的区别。
A digital referee is great, but we still need the human players to play the game and create the strategies.
数字裁判固然很好，但我们仍然需要人类选手去参加比赛并制定策略。
So if AI can be a useful tool, why are these fields medalists still so worried?
那么，如果人工智能可以成为一个有用的工具，为什么这些菲尔兹奖得主仍然如此担忧呢？
What is the worst case scenario?
最坏的情景是什么？
There are a few big concerns, Emily.
埃米莉，主要存在几点担忧。
First, there is the threat to open science.
首先，是对开放科学的威胁。
Right now, mathematics is very open.
目前，数学是非常开放的。
Researchers from all over the world share their work freely on websites like Archive.
来自世界各地的研究人员在 Archive 等网站上免费分享他们的成果。
But if AI companies take this free data, put it behind an expensive paywall, and sell it back to us,
但如果人工智能公司拿走这些免费数据，将其置于昂贵的付费墙之后，再卖还给我们，
only rich universities will be able to afford these advanced tools.
将只有富有的高校才能负担得起这些先进的工具。
This will create a huge inequality in global science.
这将在全球科学界造成巨大的不平等。
Second, there is the risk of a peer review crisis.
其次，存在同行评审危机的风险。
If AI can write thousands of mathematical papers in a few minutes, human scientists will be
如果人工智能能在几分钟内撰写数千篇数学论文，人类科学家将会
overwhelmed trying to check them all.
在试图全部核查时感到不堪重负。
The quality of scientific publishing could drop drastically.
科学出版物的质量可能会急剧下降。
That sounds like a serious challenge for the future of scientific research.
这听起来对科学研究的未来是一个严重的挑战。
It could really slow down real progress if we get flooded with low -quality, AI -generated papers.
如果我们被大量低质量、人工智能生成的论 文所充斥，这确实会拉慢真正的进步速度。
What about education?
那教育方面呢？
How will this impact young students learning math today?
这将如何影响今天学习数学的年轻学生？
Will they just use AI to do their homework?
他们会不会直接用人工智能来做作业？
That is a question many parents and teachers are asking.
这是许多家长和老师都在询问的问题。
It is very similar to when the electronic calculator was invented.
这与当年电子计算器被发明时非常相似。
Back then, people worried that students would forget how to do basic math.
那时，人们担心学生会忘记如何做基础数学。
But instead, calculators allowed us to stop spending hours on simple arithmetic and focus on higher
但相反，计算器让我们不再需要花费数小时在简单的算术上，从而专注于更高
-level mathematical concepts.
层次的数学概念。
AI should be used in the same way.
人工智能也应该以同样的方式被使用。
We must teach students how to use AI as a partner, not as a shortcut.
我们必须教会学生如何将人工智能用作合作伙伴，而不是当作捷径。
They still need to learn the basic logic of math.
他们仍然需要学习数学的基础逻辑。
Because if you do not know the basics, you won't even know if the AI is giving you a wrong answer.
因为如果你不了解基础，你甚至不会知道人工智能是否给出了错误的答案。
AI makes mistakes, too.
人工智能也会犯错误。
We call these mistakes hallucinations.
我们把这些错误称为幻觉。
Ah, yes, hallucinations.
啊，是的，幻觉。
That is when the AI confidently gives an answer that is completely made up.
那就是当人工智能自信地给出完全捏造的答案时。
In math, a tiny mistake can ruin the whole proof.
在数学中，一个微小的错误就能毁掉整个证明。
So a hallucinating AI is very dangerous. Exactly.
所以一个产生幻觉的人工智能是非常危险的。一点没错。
In math, you cannot be 99 % correct.
在数学中，你不能只有 99% 的正确率。
You are either correct or you are not.
你要么正确，要么不正确。
That is why human oversight is so critical.
这就是为什么人类的监督如此关键。
We need what we call human -in -the -loop systems, where humans and AI work together, but humans make
我们需要所谓的“人在回路”系统，在其中人类与人工智能协同工作，但由人类作出
the final decisions.
最终的决定。
So, Charles, what is the path forward?
那么，查尔斯，前进的方向是什么？
What does this group of 25 Fields Medalists want AI companies to do?
这 25 位菲尔兹奖得主希望人工智能公司做什么？
They are calling for three main things.
他们正在呼吁三件主要事项。
Transparency, respect, and collaboration.
透明度、尊重与合作。
First, AI companies must be honest about what data they are using to train their models.
首先，人工智能公司必须对其用于训练模型的规则数据保持坦诚。
Second, they need to respect intellectual property and find ways to share the benefits with the
其次，他们需要尊重知识产权，并找到与
scientific community.
科学界分享利益的方法。
And third, they must work with mathematicians to design tools that truly support research, rather
第三，他们必须与数学家合作，设计出真正支持研究的工具，而不是
than just trying to replace human scientists for commercial gain.
仅仅为了商业利益而试图替代人类科学家。
We need to build a future where AI and humanity work in harmony.
我们需要构建一个人工智能与人类和谐共存的未来。
Well, that sounds like a very reasonable and necessary request.
嗯，这听起来是一个非常合理的且必要的诉求。
Let us hope the big tech companies are listening.
希望科技巨头们正在倾听。
Charles, thank you so much for joining us today and making this complex topic so clear and
查尔斯，非常感谢你今天加入我们，把这个复杂的话题解释得如此清晰且
fascinating for our listeners.
对我们的听众来说如此引人入胜。
Thank you for having me, Emily.
感谢邀请我，艾米莉。
It was a wonderful conversation.
这是一次精彩的对话。
And a big thank you to all of our listeners for spending your time with us today on TalkCast.
也非常感谢我们所有的听众今天抽空在 TalkCast 与我们度过。
Today, we discussed how artificial intelligence is changing the world of mathematics and why top
今天，我们讨论了人工智能如何改变数学界，以及为什么顶尖
scientists are warning us about the future.
科学家们正在就未来向我们提出警告。
Now we want to hear from you.
现在我们想听听你们的想法。
Here's our question for today.
这是我们今天出的问题。
Do you think artificial intelligence will eventually become smarter than the human mind at solving
你认为人工智能最终会在解答
mathematical mysteries?
数学奥秘方面比人类大脑更聪明吗？
Or is human intuition something that can never be copied by a machine?
还是说人类的直觉 是机器永远无法复制的？
Please write down your answers in the comment section below or check out what other listeners are saying.
请在下面的评论区写下您的答案，或者看看其他听众是怎么说的。
We love reading your thoughts.
我们很乐意阅读你们的想法。
If you enjoyed this conversation, please subscribe to TalkCast.
如果您喜欢这次对话，请订阅 TalkCast。
Leave us a review and share this episode with your friends and family.
给我们留个评价，并把本集分享给您的亲朋好友。
Your feedback and support mean the world to us.
您的反馈和支持对我们至关重要。
Also, a quick reminder, the complete English transcript and audio of this episode will be posted
另外顺便提醒一下，本集的完整英文逐字稿和音频将被发布
regularly on our website at www .talkcast .top and on our WeChat official account, which is also
定期在我们的网站 www.talkcast.top 和微信公众号上，公众号也
called TalkCast.
叫作 TalkCast。
Please visit them if you want to practice your English reading and listening.
如果您想练习英语阅读和听力，请访问它们。
Until next time, keep learning, keep listening, and keep asking questions.
下次再见，保持学习，保持倾听，并保持提出问题。
I'm Emily, and this is TalkCast. Goodbye. Bye.
我是艾米莉，这里是 TalkCast。再见。拜拜。
