And without a doubt, somewhere in there are biases.

而毫無疑問，這裡面的某個環節是有偏差的。

Biases about women in the workplace.

工作場所對婦女的偏見；

Biases about people of different races.

對不同種族的人有偏見。

And all these biases are things

而所有這些偏見都是東西

companies like Google, Starbucks, and even Insider

谷歌、星巴克、甚至Insider等公司。

have been trying to train out of their workforce for years.

多年來一直在努力培養出自己的勞動力。

But is it working?

但這有用嗎？

Calvin Lai: We don't know. We really don't know.

Calvin Lai我們不知道，我們真的不知道。我們真的不知道。

Narrator: And a lot of that uncertainty

旁白：很多不確定因素

has to do with our understanding

與我們的理解有關

of implicit bias in general.

的隱性偏見一般。

So, what do we know?

那麼，我們知道什麼？

Implicit biases are stereotypes

隱性偏見是指定型觀念。

that we form about certain groups of people.

我們對某些群體形成的。

Every day, your amygdala processes

每天，你的杏仁核都會處理

and categorizes billions of stimuli.

並將數十億的刺激物進行分類。

And because of this, almost all of it

正因為如此，幾乎所有的

is happening automatically.

正在自動發生。

Every experience you have, every bit of socialization,

你的每一段經歷，每一點社交。

every piece of media is processed and stored

每一塊媒體都經過處理和存儲

by a stereotyping neural network,

由定型神經網絡。

creating a database of shortcuts

創建快捷鍵數據庫

that your brain can pull from.

你的大腦可以從。

And while it's helpful to see a big metal box with wheels

雖然看到一個帶輪子的大金屬盒子很有幫助

and immediately know it's a car,

並立即知道這是一輛車。

categorizing people in the same way is harmful.

把人歸為一類是有害的。

And that's where the implicit association test comes in.

而這就是隱性關聯測試的作用。

The IAT is part of many implicit bias trainings.

IAT是許多隱性偏見培訓的一部分。

It's supposed to establish a baseline

它應該建立一個基線

for each participant's unconscious biases by measuring them.

通過測量每個參與者的無意識偏見。

The problem is, we don't actually know if it does that.

問題是，我們其實並不知道它是否會這樣做。

Lai: I think that the implicit association test

賴：我認為隱性關聯測試

is an imperfect measure.

是一種不完善的措施。

It's been incredibly useful for researchers

它對研究人員來說非常有用

to understand how the mind works

以瞭解心智的運作

in ways that are not self-reported.

以非自我報告的方式。

But if you want to use it as a diagnostic measure

但是，如果你想把它作為一個診斷的措施

of how racist or sexist you are or something like that,

你是多麼的種族主義或性別歧視或類似的東西。

it's not gonna tell you that.

它不會告訴你。

Narrator: A psychological test is usually measured

旁白：心理測試通常是測量

in two ways: reliability and validity.

在兩個方面：可靠性和有效性。

Test-retest reliability means that people should be able

測試-複試的可靠性是指人們應該能

to take the test over and over again

屢試不爽

and get nearly the same results each time.

並每次得到幾乎相同的結果。

A perfect reliability score is a one,

可靠性滿分是一。

but a test is solid if it scores at least 0.7.

但如果考試成績在0.7分以上，就是實心的。

But studies have put the race IAT's reliability at 0.44

但研究認為，比賽IAT的可靠性為0.44

and the IAT overall at 0.5, well below acceptable standards.

和IAT整體為0.5，遠低於可接受的標準。

This means when a person takes the test multiple times,

這意味著當一個人多次參加考試。

they get notably different results.

他們得到的結果明顯不同。

And experts can't say for sure

而專家們也不能確定

whether that's because the test is a bad tool

是否因為測試是一個糟糕的工具？

or if the concept of implicit bias actually fluctuates.

或隱性偏見的概念是否真的有波動。

Lai: So, one of the theoretical ambiguities right now

賴。所以，現在理論上的一個模糊地帶

about the nature of implicit bias

關於隱性偏見的性質

is the extent to which it reflects something deep

是指它在多大程度上反映了一些深層次的

about a personality or an enduring attitude

風骨

versus something that is more fickle,

與更易變的東西。

something more like an emotional state or a mood.

更像是一種情緒狀態或心情。

Narrator: Either way, this discrepancy

旁白：無論如何，這種差異

makes it difficult to rely on the results of the test.

使得檢測結果難以靠譜。

Which leads us to validity:

這就導致了我們的有效性。

Does the test actually predict what it says it does?

這個測試是否真的能預測到它所說的作用？

The IAT was intended to and is currently used in a way

獨立行政法庭的目的和目前的使用方式是：

that is supposed to link implicit bias to behaviors.

應該是把隱性偏見和行為聯繫起來的。

At least four different meta-analyses

至少有四項不同的元分析。

between 2007 and 2015 looked at this exact thing.

2007年到2015年之間，看的就是這個事情。

And all of them suggest that the IAT

而所有這些都表明，IAT

doesn't really predict behaviors that well.

並不能很好地預測行為。

Lai: It turns out that predicting discrimination

賴：原來，預測歧視

is just difficult, full stop.

就是很難，完全停止。

Narrator: But even if the IAT was conclusively valid

旁白：但即使IAT是絕對有效的... ...

and reliable, implicit bias trainings still have a problem:

和可靠的，隱性偏差培訓還是有問題的。

Acknowledging your bias does not mean

承認自己的偏見並不意味著......。

you're gonna act less racist.

你要去行動 少種族主義。

At least one study found that recognizing a bias

至少有一項研究發現，承認偏見

is a necessary step to getting rid of it,

是擺脫它的必要步驟。

but it won't solve the problem on its own.

但它本身並不能解決這個問題。

Lai: You want to make people feel enough motivation,

賴：你要讓大家感受到足夠的動力。

maybe something like enough guilt or shame

也許是足夠的內疚或羞愧

to be actually motivated to do something

奮發有為

about the problem of bias or discrimination.

關於偏見或歧視的問題；

Narrator: And some studies suggest that the trainings

旁白：一些研究表明

can even make the problem worse.

甚至會使問題更加嚴重。

They have the potential to dredge up stereotypes

他們有可能挖出一些陳舊的觀念。

and make a person act on them more.

並使人更加行動起來。

So, how do we make sure that

那麼，我們如何確保

implicit bias trainings are effective?

隱性偏見培訓有效嗎？

Standardizing the programs is a start.

標準化的方案是一個開始。

Lai: I've seen ones where the IAT is never mentioned,

賴：我見過的都是從來不提IAT的。

let alone used in any way.

更不要說用於任何方面了。

I've seen ones where they put a measurement of bias

我見過有的人把測量的偏差

in the middle of the thing,

中的事情。

and they do it as a group activity.

他們把它作為一個團體活動來做。

Some of them are just purely just a collection

有的只是單純的收藏而已

of PowerPoint slides with nothing else attached.

的PowerPoint幻燈片，沒有任何其他附件。

Narrator: One 2016 meta-analysis

敘述者：2016年的一項元分析

of over 40 years of diversity-training data

40多年的多樣性培訓數據

found that the programs were successful

發現這些方案是成功的

if they focused on skill development

如果他們專注於技能發展

and were conducted over a long period of time.

並進行了長時間的。

Many trainings end up being reactionary,

很多培訓最後都是反動的。

half-day or shorter events that end up

半天或更短的活動，最後

being more performative than impactful.

更多的是表現而不是影響。

It's like using a Band-Aid to treat a broken bone

就像用Ok繃來治療骨折一樣

instead of a comprehensive treatment plan.

而不是一個全面的治療計劃。

Lai: Exactly!

阿來就是這樣！

You might not be so certain

你可能不那麼肯定

that any individual initiative works,

任何個人的舉措都是有效的。

but we know that, generally,

但我們知道，一般來說，。

when firms or companies have more of them,

當企業或公司擁有更多。

their diversity and inclusion outcomes tend to be better.

其多樣性和包容性成果往往更好。

Narrator: Harvard Business Review

旁白：哈佛商業評論

found that a combination of things

發現

like college recruitment, mentoring programs,

比如高校招生、導師計劃。

self-managed teams, and task forces

自我管理的團隊和工作隊

have increased diversity.

增加了多樣性。

Diversity task forces alone

僅僅是多樣性工作隊

boosted Black women in management by 23%.

將黑人婦女在管理層的地位提高了23%。

Instead of focusing on the thoughts of individual employees,

而不是關注員工個人的想法。

these tactics address the bigger, systemic issues

這些策略可以解決更大的系統性問題。

that make bias a problem in the first place.

使得偏見首先成為一個問題。

So, just because we don't know for sure

所以，只是因為我們不確定

if implicit bias trainings work doesn't mean they can't.

如果隱性偏見培訓有效，並不意味著他們不能。

We should be asking ourselves

我們應該問自己

what we do know we can do differently

知易行難

in order to ensure lasting change.

以確保持久的變化。

I just want to be very clear that the information

我只是想非常清楚地表明，這些資訊。

we're presenting here has to do with implicit bias trainings

我們在這裡介紹的是關於隱性偏見的培訓。

and their effect on diversity

及其對多樣性的影響

and inclusion within companies.

和公司內部的包容。

Bias is very much real, and it's something

偏見是非常真實的，它的東西。

that we as individuals need to continue

我們作為個人需要繼續

being aware of, checking, and unlearning.

覺察、檢查、解除學習。

So, please, continue the conversation in the comments below

所以，請大家在下面的評論中繼續對話。

and in real life, and subscribe for more "Deep Science."

並在現實生活中，並訂閱更多 "深度科學"。