How AI and Democracy Can Fix Each Other | Divya Siddarth | TED

翻译人员: Lening Xu 校对人员: Yip Yan Yeung

Recently I told someone my work is on democracy and technology.

最近我告诉一个人， 我的工作是关于民主和科技的。

He turned to me and said,

他转向我说：

“Wow...

“哇……

I’m sorry.”

什么？”

(Laughter)

（笑声）

But I love my work.

但我喜欢我的工作。

I know that we can build a world

我知道我们可以建设一个世界，

where technological marvels are directed towards people's benefit,

投入资源让科技奇迹服务于人类福祉。

using their input.

We have gotten so used to seeing democracy as a problem to be solved.

我们已经习惯于将民主 视为一个需要解决的问题。

But I see democracy as a solution,

但我将民主视为解决方案，

not as a problem.

而不是问题。

Democracy was once a radical political project,

民主曾是一个激进的政治项目，

itself a cutting-edge social technology,

它本身就是一种尖端的社会技术，

a new way to answer the very question we are faced with now,

是回答我们当下 所面临问题的一种新方法，

in the era of artificial intelligence:

当下的人工智能时代：

how do we use our capabilities to live well together?

如何利用我们的能力 更好地共同生活。

We are told that transformative technologies like AI are too complicated,

有人告诉我们， 像人工智能这样的变革性技术过于复杂，

or too risky

风险太大，

or too important to be governed democratically.

或太重要， 无法通过民主方式进行治理。

But this is precisely why they must be.

但这正是民主必须存在的原因。

If existing democracy is unequal to the task,

如果现有民主 无法胜任这项任务，

our job is not to give up on it.

我们的工作就不是放弃它。

Our job is to evolve it.

我们的工作是对其进行改进。

And to use technology as an asset to help us do so.

并使用技术作为资产 来帮助我们做到这一点。

Still, I understand his doubts.

不过，我理解他的疑虑。

I never meant to build my life around new forms of democracy.

我从来没有想过要 围绕新的民主形式建立自己的生活。

I started out just really believing in the power of science.

一开始我真的相信科学的力量。

I was modifying DNA in my kitchen at 12,

我 12 岁时正在厨房里修改DNA，

and when I got to Stanford as a computational biology major,

当我来到斯坦福大学 攻读计算生物学时，

I was converted to a new belief -- technology.

我皈依了一种新的信仰——科技。

I truly believed in the power of tech to change the world.

我真的相信 科技改变世界的力量。

Maybe, like many of you.

也许就像你们中的许多人一样。

But I saw that the technologies that really made a difference

但我看到，真正发挥作用的技术，

were the ones that were built with and for the collective.

是那些与集体合作、 为集体开发的技术。

Not the billions of dollars

不是把数十亿美元投入 第 19 款助长成瘾的社交应用。

pumped into the 19th addiction-fueling social app.

But the projects that combine creating something truly new

而是这些项目， 它们将创造真正的新事物，

with building in ways for people to access,

与以供人们访问、 受益和指导的方式相结合。

benefit from and direct it.

Instead of social media, think of the internet.

代替社交媒体，想想互联网。

Built with public resources on open standards.

以开放标准为基础，利用公共资源构建。

This is what brought me to democracy.

这就是我走向民主的原因。

Technology expands what we are capable of.

技术扩展了我们的能力。

Democracy is how we decide what to do with that capability.

民主是我们决定 如何利用这种能力的方式。

Since then, I've worked on using democracy as a solution

从那时起， 我一直努力使用民主作为解决方案，

in India, the US, the UK, Taiwan.

在印度、美国、英国、台湾，

I've worked alongside incredible collaborators

我曾与优秀的合作者合作，

to use democracy to help solve COVID,

利用民主来帮助解决新冠问题，

to help solve data rights.

帮助解决数据权限问题。

And as I'll tell you today,

正如我今天要告诉你的那样，

to help solve AI governance with policymakers around the world

与世界各地的决策者以及 OpenAI和Anthropic等尖端科技公司。

and cutting-edge technology companies like OpenAI and Anthropic.

一起帮助解决人工智能治理问题。

How?

怎么做呢？

By recognizing that democracy is still in its infancy.

承认民主仍处于起步阶段。

It is an early form of collective intelligence,

这是集体智慧的早期形式，

a way to put together decentralized input from diverse sources

是一种汇集来自不同来源的分散输入，

and produce decisions that are better than the sum of their parts.

并做出比简单合计更好决策的方式。

That’s why, when my fantastic cofounder Saffron Huang and I

这就是为什么，当我和出色的 联合创始人黄珊（Saffron Huang），

left our jobs at Google DeepMind and Microsoft

离开谷歌DeepMind和微软的工作，

to build new democratic governance models for transformative tech,

去为变革性科技建立 新的民主治理模式时，

I named our nonprofit the Collective Intelligence Project,

我将我们的非营利组织 命名为集体智慧项目（CIP），

as a nod to the ever-evolving project of building collective intelligence

以此致敬，

那些不断发展的建设集体智慧 以促进集体繁荣的项目。

for collective flourishing.

Since then we've done just that,

从那时起，我们就这样做了，

building new collective intelligence models to direct artificial intelligence,

建立了新的集体智能模型 来指导人工智能，

to run democratic processes.

运行民主进程。

And we've incorporated the voices of thousands of people into AI governance.

而且我们已经将成千上万人的意见 纳入了人工智能治理。

Here are a few of the things we've learned.

以下是我们学到的一些东西。

First, people are willing and able to have difficult,

首先，人们愿意并且能够，

就细微差别的话题 进行艰难、复杂的对话。

complex conversations on nuanced topics.

When we asked people about the risks of AI they were most concerned about,

当我们询问人们， 他们最关心的人工智能风险时，

they didn't reach for easy answers.

他们没有随口一答。

Out of more than 100 risks put forward,

在提出的 100 多种风险中，

the top-cited one: overreliance on systems we don't understand.

最频繁出现的风险是： 过度依赖我们不了解的系统。

We talked to people across the country,

我们与全国各地的人进行了交谈，

from a veteran in the Midwest to a young teacher in the South.

从中西部的退伍军人到南方的年轻教师。

People were excited about the possibilities of this technology,

人们对这项技术的可能性感到兴奋，

but there were specific things they wanted to understand

但是他们想了解一些具体的信息，

about what models were capable of before seeing them deployed in the world.

在将模型向全世界发布之前， 了解模型能够做到一些什么。

A lot more reasonable than many of the policy conversations that we're in.

比正在进行的许多政策对话合理得多。

And importantly, we saw very little of the polarization

重要的是，我们几乎看不到 我们经常听到的两极分化。

we're always hearing about.

On average, just a few divisive statements

平均而言，数百份共识声明中 只有几份意见相左的声明。

for hundreds of consensus statements.

Even on the contentious issues of the day,

即使在当今有争议的问题上，

like free speech or race and gender,

例如言论自由、种族和性别，

we saw far more agreement than disagreement.

我们看到的共识也远多于分歧。

Almost three quarters of people agree that AI should protect free speech.

近四分之三的人同意 人工智能应该保护言论自由。

Ninety percent agree that AI should not be racist or sexist.

90% 的人同意人工智能 不应该有种族主义或性别歧视。

Only around 50 percent think that AI should be funny though,

但是，只有大约 50％ 的人 认为人工智能应该很有趣，

so they are still contentious issues out there.

因此它们仍然是有争议的问题。

These last statistics

最新的统计数据来自

are from our collective constitution project with Anthropic,

我们与Anthropic的集体宪法项目，

where we retrained one of the world's most powerful language models

我们对世界上最强大的语言模型之一 进行了再训练。

on principles written by 1,000 representative Americans.

根据 1,000 名具有代表性的 美国人撰写的原则，

Not AI developers or regulators or researchers at elite universities.

不是人工智能开发人员、监管机构 或精英大学的研究人员。

We built on a way of training AI

我们建立了一种训练人工智能的方法，

that relies on a written set of principles or a constitution,

这种方法依赖于一套成文的原则或章程，

we asked ordinary people to cowrite this constitution,

我们请普通人共同撰写这部宪法，

we compared it to a model that researchers had come up with.

并将其与研究人员 研发的模型进行了比较。

When we started this project, I wasn't sure what to expect.

当我们开始这个项目时， 我不确定会发生什么。

Maybe the naysayers were right.

也许反对者是对的。

AI is complicated.

人工智能很复杂。

Maybe people wouldn't understand what we were asking them.

也许人们不明白我们要他们做些什么。

Maybe we'd end up with something awful.

也许我们最终会遇到一些可怕的事情。

But the people’s model, trained on the cowritten constitution,

但基于共同撰写的宪法 进行训练的人类模型，

was just as capable and more fair

与研究人员开发的模型 一样有效，也更公平。

than the model the researchers had come up with.

People with little to no experience in AI

几乎没有人工智能经验的人，

did better than researchers, who work on this full-time,

在构建更公平的聊天机器人方面， 比全职研究该领域的研究人员做得更好。

in building a fairer chatbot.

Maybe I shouldn't have been surprised.

也许我不应该感到惊讶。

As one of our participants from another process said,

正如另一个项目的一位参与者所说，

"They may be experts in AI, but I have eight grandchildren.

“他们可能是人工智能专家， 但我有八个孙子。

I know how to pick good values."

我知道如何衡量正确的价值观。”

If technology expands what we are capable of

如果技术扩展了我们的能力，

and democracy is how we decide what to do with that capability,

而民主是我们决定如何 利用这种能力的方式，

here is early evidence that democracy can do a good job deciding.

这里有早期的证据表明 民主可以很好地做出决定。

Of course, these processes aren't enough.

当然，这些工序还不够。

Collective intelligence requires a broader reimagining of technology

集体智慧需要对技术和民主 进行更广泛的重新构想。

and democracy.

That’s why we’re also working on co-ownership models

这就是为什么我们还在 开发共同所有权模型，

for the data that AI is built on -- which, after all, belongs to all of us --

因为构建人工智能的数据 ， 最终属于我们所有人，

and using AI itself to create new and better decision-making processes.

利用人工智能本身来创建 新的、更好的决策流程。

Taking advantage of the things

利用语言模型来做人类无法做的事情，

that language models can do that humans can’t,

like processing huge amounts of text input.

例如处理大量的文本输入。

Our work in Taiwan has been an incredible test bed for all of this.

我们在台湾的工作 是这一切的完美试验田。

Along with Minister Audrey Tang and the Ministry of Digital Affairs,

我们与唐凤部长（Audrey Tang） 和台湾数位发展部一道，

we are working on processes to ask Taiwan's millions of citizens

制定程序并询问了台湾数百万公民，

what they actually want to see as a future with AI.

有人工智能的未来 在他们眼里到底是什么样的。

And using that input not just to legislate,

收集这些不仅是为了立法，

but to build.

而且是为了建设。

Because one thing that has already come out of these processes

因为从这些流程中浮现出的一点是，

is that people are truly excited about a public option for AI,

人们对人工智能的公共选择 感到非常兴奋，

one that is built on shared public data that is reliably safe,

这种选择建立在 可靠安全的共享公共数据之上，

that allows communities to access, benefit from

允许社区访问这些数据，从中受益，

and adjust it to their needs.

并根据自己的需求进行调整。

This is what the world of technology could look like.

这就是技术世界的样子。

Steered by the many for the many.

由众人，为众人。

I often find that we accept unnecessary trade-offs

我经常发现，在变革性技术方面， 我们接受了不必要的妥协。

when it comes to transformative tech.

We are told that we might need to sacrifice democracy

有人告诉我们， 我们可能需要牺牲民主，

for the sake of technological progress.

为了技术进步。

We have no choice but to concentrate power to keep ourselves safe

我们别无选择， 只能集中力量保护自己，

from possible risks.

免受可能的风险。

This is wrong.

这是不对的。

It is impossible to have any one of these things --

我们不可能得到 这三项里的任意一项——

progress, safety or democratic participation --

进步、安全或民主权利，

without the others.

如果这三项中缺失任一。

If we resign ourselves to only two of the three,

如果我们选择只保留三项中的两项，

we will end up with either centralized control or chaos.

我们最终将陷入集中控制或混乱。

Either a few people get to decide or no one does.

要么由少数人决定， 要么没人决定。

These are both terrible outcomes,

这两个结果都很糟糕，

and our work shows that there is another way.

我们的研究表明还有另一种方法。

Each of our projects advanced progress,

我们的每个项目都推动了 进步、安全和民主参与，

safety and democratic participation

by building cutting-edge democratic AI models,

通过建立尖端的民主人工智能模型，

by using public expertise as a way to understand diffuse risks

利用公共专业知识 作为理解分散风险的方式，

and by imagining co-ownership models for the digital commons.

以及想象数字公域的共同所有权模式。

We are so far from the best collective intelligence systems we could have.

距离能实现的最好的集体智慧系统， 我们还有很长的路要走。

If we started over on building a decision-making process for the world,

如果我们重新开始为世界 建立决策流程，

what would we choose?

我们会选择什么？

Maybe we'd be better at separating financial power from political power.

也许我们可以更好地 将金融力量与政治权力分开。

Maybe we'd create thousands of new models of corporations

也许我们会创建成千上万种 新的企业或官僚机构模式。

or bureaucracies.

Maybe we'd build in the voices of natural elements

也许我们会听取自然元素 或子孙后代的声音。

or future generations.

Here's a secret.

我要说个秘密。

In some ways, we are always starting from scratch.

在某些方面，我们总是从头开始。

New technologies usher in new paradigms

新技术带来了新范式，

that can come with new collective intelligence systems.

这可能伴随着新的集体智能系统。

We can create new ways of living well together

我们可以创造新的共同生活方式，

if we use these brief openings for change.

如果我们利用这些短暂的变革机会，

The story of technology and democracy is far from over.

科技和民主的故事远未结束。

It doesn't have to be this way.

不一定非得这样。

Things could be unimaginably better.

情况可能出人意料地好。

As the Indian author Arundhati Roy once said,

正如印度作家阿伦哈蒂·罗伊 （Arundhati Roy）曾经说过的那样，

"Another world is not only possible,

“另一个世界不仅是可能的， 而且她正在路上。

she is on her way.

On a quiet day, I can hear her breathing."

在安静的日子里，我能听见她的呼吸。”

I can hear our new world breathing.

我能听见我们的新世界在呼吸。

One in which we shift the systems we have

我们将现有的制度，

towards using the solution of democracy to build the worlds we want to see.

转向使用民主的解决方案来建设的 我们想要看到的世界。

The future is up to us.

未来取决于我们。

We have a world to win.

我们要赢得一个世界。

Thank you.

谢谢。

(Applause)

（掌声）