AI and the Paradox of Self-Replacing Workers | Madison Mohns | TED

翻译人员: Yip Yan Yeung 校对人员: Gentle Yang

I'm going about my day, normal Tuesday of meetings

我正做着日常的工作， 例行的周二会议，

when I get a ping from my manager's manager's manager.

这时我的经理的经理的经理 给我发来了一条消息，

It says: “Get me a document by the end of the day

上面写着： 今天结束之前，交给我一份文件，

that records everything your team has been working on related to AI."

介绍下你的团队做的 所有与 AI 相关的工作。

As it turns out, the board of directors of my large company

事实证明，我所在的 这家大公司的董事会成员们，

had been hearing buzz about this new thing called ChatGPT,

一直听到有关 ChatGPT 这个新玩意的热议，

and they wanted to know what we were doing about it.

他们想知道我们 在做一些什么和它有关的工作。

They are freaking out about the future,

他们对未来十分慌张，

I'm freaking out about this measly document,

而我也对这份微不足道的文档十分慌张，

it sounds like the perfect start

听起来就像是解决科技届下一个 最火爆的问题的完美起点，对吧？

to solving the next hottest problem in tech, right?

As someone who works with machine-learning models

作为一个每天都在 使用机器学习模型的人，

every single day,

I know firsthand that the rapid development of these technologies

我亲身体会到这些技术的快速发展

poses endless opportunities for innovation.

为创新带来了无限的机会。

However, the same exponential improvement in AI systems

然而，人工智能系统也有同样的指数级改进

is becoming a looming existential threat to the team I manage.

这对我的团队来说已是迫在眉睫的生存威胁。

With increasing accessibility

随着 AI 研究领域越来越普及，

and creepily human-like results coming out of the field of AI research,

生成了令人毛骨悚然的 类似人类可以产生的结果，

companies like my own are turning toward automation to make things more efficient.

像我所在的公司就会 转而使用自动化，达成更高的效率。

Now on the surface, this seems like a pretty great vision.

从表面上看， 这似乎是一个非常美好的愿景。

But as we start to dig deeper, we uncover an uncomfortable paradox.

但是，当我们开始深入挖掘时， 我们发现了一个令人不安的矛盾。

Let's break this down.

我们来详细分析一下。

In order to harness the power of AI systems,

为了利用 AI 系统的力量，

these systems must be trained and fine-tuned

为符合高质量标准， 这些系统必须接受训练和微调。

to match a high-quality standard.

But who defines quality,

但是，话说回来，由谁来定义“质量”？ 由谁去训练这些模型？

and who trains these systems in the first place?

As you may have guessed, real-life subject matter experts,

正如你可能已经猜到的那样， 是真实的领域专家，

oftentimes the same exact people who are currently doing the job.

通常也就是进行这项工作的同一批人。

Imagine my predicament here.

想象一下我在这里的困境。

I get to go to my trusted team, whom I've worked with for years,

我要去找我信任的团队， 我已经与他们共事多年，

look them in the eyes

直视他们的眼睛，

and pitch them on training the very systems that might displace them.

说服他们训练 那些将要取代他们的系统。

This paradox had led me to rely on three ethical principles

这个矛盾让我依赖于 三条道德原则，

that can ensure that managers can grapple with the implications

这些原则可以确保 管理者能够应对自我取代的劳动力

of a self-replacing workforce.

带来的影响。

One, transformational transparency,

第一，变革的透明度；

Two, collaborative AI augmentation.

第二，协作式 AI 增强；

And three, reskilling to realize potential.

第三，通过学习新技能发掘潜力。

Now before we get into solutions, let’s zoom out a little bit.

在我们讨论解决方案之前， 让我们退后一步来看。

How deep is this problem of self-replacing workers, really?

自我取代的员工问题 到底有多严重？

Research from this year coming out of OpenAI

OpenAI 今年发布的研究表明，

indicates that approximately 80 percent of the US workforce

大约 80％ 的美国员工

could see up to 10 percent of their tasks impacted

可以预见 AI 的到来会造成 他们高达 10% 的工作任务受到影响，

by the introduction of AI,

while around 19 percent of the workforce

而大约 19％ 的员工

could see up to 50 percent of their tasks impacted.

可以预见高达 50% 的工作 将受到影响。

The craziest thing about all of this is,

最可怕的是， 这些技术一视同仁。

is that these technologies do not discriminate.

Occupations that have historically required an immense amount of training

长久以来一直需要大量培训

or education are equally as vulnerable to being outsourced to AI.

或教育的岗位同样 容易受到被 AI 取代的威胁。

Now before we throw our hands up and let the robots take over,

在我们彻底摆烂， 让机器人接手一切之前，

let's put this all into perspective.

让我们客观地分析一下。

Fortunately for us,

所幸对我们来说，

this is not the first time in history that this has happened.

这不是历史上第一次发生这种情况。

Let's go back to the Industrial revolution.

让我们回到工业革命。

Picture Henry Ford’s iconic Model T automobile production line.

想象一下亨利·福特（Henry Ford） 标志性的 Model T 汽车生产线。

In this remarkable setup,

在这个卓越的系统中，

workers and machines engage in a synchronous dance.

工人和机器能一唱一和。

They were tasked with specific repetitive tasks,

他们的任务是执行特定的重复性任务，

such as tightening bolts or fitting components

比如在流水线上拧紧螺丝或安装零件。

as the product moved down the line.

Ironically, and not dissimilar to my current predicament,

讽刺且无异于我现处的困境的是，

the humans themselves played a crucial role in training the systems

人类本身在训练系统过程中 发挥了至关重要的作用，

that would eventually replace their once multi-skilled roles.

而这些系统最终将取代 他们曾经的多技能岗位。

They were the ones who honed their craft, perfected the techniques

他们磨练了自己的手艺， 完善了技术，

and ultimately handed off the knowledge to the technicians

最终将知识交给了 参与自动化整个流程的

and engineers involved in automating their entire process.

技术人员和工程师。

Now on the outset, this situation seems pretty dire.

一开始，局面似乎非常严峻。

Yet despite initial fears and hesitations

然而，尽管这些技术进步 最初引起了恐惧和犹豫，

involved in these technological advancements,

history has proven that humans have continuously found ways

但历史已经证明， 人类一直在寻找

to adapt and innovate.

适应和创新的方法。

While some roles were indeed replaced, new roles emerged,

虽然有些岗位确实被取代了， 但出现了新的岗位，

requiring higher-level skills like creativity

需要更高级别的技能，例如创意

and creative problem solving that machines just simply couldn't replicate.

和创造性解决问题的能力， 而这些技能是机器根本无法复制的。

Reflecting on this historical example

反思这个先例，

reminds us that the relationship between humans and machines

它提醒了我们， 人与机器之间的关系

has always been a delicate balancing act.

一直是一种微妙的平衡。

We are the architects of our own progress,

我们是自己进步的建筑师，

often training machines to replace us

经常训练机器取代我们，

while simultaneously carving out unique roles for ourselves

同时为自己塑造了独特的角色，

and discovering new possibilities.

发掘了新的可能性。

Now coming back to the present day, we are on the cusp of the AI revolution.

回到今天，我们正处于 AI 革命的风口浪尖。

As someone responsible for moving that revolution forward,

作为负责推动这场革命的人，

the tension becomes omnipresent.

紧张局势变得无处不在。

Option one, I can innovate quickly and risk displacing my team.

选项一，我可以快速创新， 冒着取代团队的风险。

Or option two, I can refuse to innovate in an effort to protect my team,

或者选择二，我可以 为了保护我的团队而拒绝创新，

but ultimately still lose people because the company falls behind.

但最终还是会因为公司落后 而失去员工。

So what am I supposed to do

那么，在这种情况下，

as a mere middle manager in this situation?

作为一名中层管理者， 我该怎么做呢？

Knowingly introducing this complex paradox for your team

有意为你的团队引入 这种复杂的矛盾

presents strong challenges for people management.

给人员管理带来了严峻的挑战。

Luckily, we can refer back to those three ethical principles

幸运的是，我们可以回顾 我在演讲开始时提到的三条道德原则，

I addressed at the beginning of the talk

to ensure that you can continue to move ahead

确保你能够在不让员工掉队的情况下 继续向前迈进。

without leaving your people behind.

First and foremost,

首先，

AI transformation needs to be transparent.

AI 的转型需要保持透明。

As leaders, it is imperative to foster dialogue,

作为领导者， 当务之急就是促进对话，

address key concerns,

解决关键问题，

and offer concise explanations regarding the purpose

并就实施 AI 的目的和潜在挑战 提供简明的解释。

and potential challenges entailed in implementing AI.

This requires actively involving your employees

这需要积极地让您的员工

in the decision-making process

参与决策过程

and valuing their autonomy.

并重视他们的自主权。

By introducing the concept of consent,

通过引入“同意”的概念，

especially for employees who are tasked

尤其是那些被指派自动化 自己核心工作内容的员工，

with automating their core responsibilities,

we can ensure that they maintain a strong voice

我们可以确保他们在 塑造自己的职业命运方面

in carving out their professional destiny.

拥有强有力的发言权。

Next, now that we've gotten folks bought into this grandiose vision

接下来，既然我们已经让 人们接受了这个宏大的愿景，

while acknowledging the journey that lies ahead,

同时也认可了未来的旅程，

let's talk about how to use AI as an augmentation device.

让我们来谈谈如何 将 AI 用作增强工具。

Picture the worst part of your job today.

想象一下你今天工作中最糟糕的部分。

What if you could delegate it?

如果你能委托于他人，怎么样？

And no, not hand it off to some other sad soul at work,

不，不要把它丢给 其他悲惨的打工人，

but hand it to a system that can do your rote tasks for you.

而是把它交给一个可以 替你完成机械性任务的系统。

Instead of perceiving AI as a complete replacement,

与其将 AI 视为完全的替代品，

identify opportunities where you can use it

不如找出可以利用它 提高员工潜力和生产力的机会。

to enhance your employees' potential and productivity.

Collaboratively with your team,

与你的团队合作，

identify areas and tasks that can be automated,

确定可以被自动化的领域和任务，

carving out more room for higher-value activities

为需要批判性思维的 高价值活动腾出更多空间，

requiring critical thinking that machines just aren't very good at doing.

而机器并不擅长这类活动。

Let's put this into an example.

举个例子。

Recently, I completed a project with my team at work

最近，我与我的团队 一起完成了一个项目，

that's going to save our company over 12,000 working hours.

这将为我们公司节省 超过 12000 个工时。

The folks involved in training this algorithm

参与训练该算法的人员

are the same subject matter experts that worked tirelessly last year

就是同一批领域专家， 他们去年夜以继日地

to hand-curate and research data to optimize segmented experiences

精心审核、研究数据， 优化我们网站各个部分的体验。

across our website.

Now because of the sheer amount of time spent and the level of detail involved,

由于花费了大量时间、 涉及了大量细节，

I would have expected

我本以为这个任务 会让人感到非常自豪。

that there was an immense amount of pride behind this workflow.

But to my surprise, as it turns out,

但令我惊讶的是，事实证明，

the subject matter experts who built this model

构建这个模型的领域专家

were actually excited to hand these tasks off to automation.

其实很高兴将这些任务交给自动化。

There were things that they would have much rather spent their time on,

他们更愿意把时间花在一些事情上，

like in optimizing existing data to perform better on product surfaces

比如优化现有数据， 改善产品层面的表现，

or even researching and developing new insights to augment

甚至是研究和开发新的洞察，

where the model just simply doesn't do as well.

增强模型的薄弱点。

Lastly, we must reskill in order to avoid replacement.

最后，我们必须掌握新技能， 避免被取代。

Knowingly investing in the professional development

有意识地为员工的 职业发展和福祉投资，

and well-being of our workforce

ensures that they are equipped with the skills and knowledge

确保他们具备在 AI 驱动的未来 蓬勃发展所需的技能和知识。

needed to thrive in an AI-powered future.

By providing opportunities for upskilling and reskilling,

通过提供提高技能 和学习新技能的机会，

we can empower our employees to rethink their roles as they exist today

我们可以使我们的员工 重新思考他们现在所在的岗位，

and carve out new possibilities that align with their evolving expertise

并根据他们不断变化的专业知识和兴趣 开拓新的可能性。

and interests.

So how does this work in practice?

那么该如何实践呢？

When I started introducing AI as a way to accelerate my team's workflows,

当我开始引入 AI 作为加快团队 工作流程的一种方式时，

I used it as an opportunity to improve my team's technical literacy.

我把它当作提高团队技术素养的机会。

I worked with my team of engineers on a tool

我与我的工程师团队 合作开发了一款工具，

that could transparently identify

可以透明地识别 数据对模型结果的影响。

the impact of data on a model's outcomes.

I then went to my operations analyst,

然后我去找了我的运营分析师，

who didn't have technical training at the time,

他/她当时没有接受过技术培训，

and they were able to quickly identify areas where the model was underperforming

但能够快速确定模型 表现不佳的领域，

and hand off direct suggestions to my data science team

并直接向我的数据科学团队提出建议，

to make those models do better next time.

让这些模型下次表现得更好。

Fostering a culture of continuous learning

培养持续学习 和学习新技能的文化至关重要。

and reskilling is paramount.

It makes AI transformation a lot more exciting and a lot less scary.

它使 AI 转型 更加令人兴奋，更不令人害怕。

We have reached a critical juncture

我们已经到了一个关键时刻，

where the rapid development of AI technology

AI 技术的快速发展

poses both opportunities and challenges.

带来了机遇和挑战。

As managers and leaders,

作为管理者和领导者，

it is imperative that we navigate this terrain

我们必须以敏锐和远见 驾驭这一领域。

with both sensitivity and foresight.

By embracing innovation,

通过拥抱创新、 培养具有适应性的文化，

fostering a culture of adaptation,

and ultimately intentionally investing in the professional development

有意地为员工的职业发展和福祉投入，

and well-being of our workforce,

we can ensure that we are preparing our team

我们可以确保 我们在为我们的团队

for the challenges that lie ahead

为迎接未来的挑战做好准备，

while addressing the complexities of introducing AI.

同时解决引入 AI 所带来的复杂性。

Together, let's forge a future that harmoniously combines human ingenuity

让我们共同打造一个将人类智慧

和技术进步和谐结合的未来，

and technological progress,

where AI enhances human potential

届时 AI 可以增强

rather than replacing it.

而不是取代人类的潜能。

Thank you.

谢谢。

(Applause)

（掌声）