Can AI predict someone's breakup? - Thomas Hofweber

譯者: Lilian Chiu 審譯者: 麗玲 辛

You and your partner Alex have been in a strong, loving relationship for years,

你和你的伴侶艾力克斯多年來 維持著穩固、相愛的關係，

and lately you're considering getting engaged.

最近你們在考慮訂婚。

Alex is enthusiastic about the idea, but you can’t get over the statistics.

亞力克斯對這個想法很熱衷， 但你卻無法忘懷統計數字。

You know a lot of marriages end in divorce, often not amicably.

你知道很多婚姻是離婚收場， 且通常是不歡而散。

And over 10% of couples in their first marriage get divorced

且初次結婚的夫妻中有超過 10%

在結婚的前五年內就會離婚。

within the first five years.

If your marriage wouldn’t even last five years,

如果你們的婚姻連五年都撐不到，

you feel like tying the knot would be a mistake.

你會認為結婚是個錯誤。

But you live in the near future,

但你生活在不久的將來，

where a brand-new company just released an AI-based model

一間全新的公司剛剛推出了 一個人工智慧模型，

that can predict your likelihood of divorce.

可以預測你離婚的可能性。

The model is trained on data sets

訓練模型用的資料集

containing individuals’ social media activity,

包括個人的社群媒體活動、

online search histories, spending habits, and history of marriage and divorce.

線上搜尋歷史記錄、花費習慣，

以及婚姻與離婚史。

And using this information,

人工智慧可以使用這些資訊來預測

the AI can predict if a couple will divorce

夫妻是否會在結婚後的五年內離婚，

within the first five years of marriage with 95% accuracy.

正確率為 95%。

The only catch is the model doesn’t offer any reasons for its results—

唯一的問題是，這個模型不會 說明它的結果是怎麼得到的。

it simply predicts that you will or won’t divorce without saying why.

它只會預測你是否會離婚， 不會解釋為什麼。

So, should you decide whether or not to get married

那麼，你是否該根據人工智慧的 預測來決定是否要結婚？

based on this AI’s prediction?

Suppose the model predicts you and Alex

假設模型預測你和艾力克斯 在結婚後的五年內會離婚，

would divorce within five years of getting married.

At this point, you'd have three options.

此時，你有三個選擇。

You could get married anyway and hope the prediction is wrong.

你可以不管預測，還是結婚， 並希望預測是錯的。

You could break up now,

你可以現在分手，

though there’s no way to know if ending your currently happy relationship

不過就永遠無法知道 結束你當前幸福的關係

would cause more harm than letting the prediction run its course.

造成的傷害是否會比 讓預測自然發展來得更大。

Or, you could stay together and remain unmarried,

或者你們可以繼續 在一起，不要結婚，

on the off-chance marriage itself would be the problem.

以免婚姻本身就是問題。

Though without understanding the reasons for your predicted divorce,

不過，因為無法知道 預測你會離婚的原因，

you’d never know if those mystery issues would still emerge

你永遠不會知道這些未知的 問題將來是否仍然會浮現，

to ruin your relationship.

毀掉你們的關係。

The uncertainty undermining all these options

所有這些選項背後的不確定性

stems from a well known issue with AI around explainability and transparency.

都來自於人工智慧 一個眾所皆知的問題：

可解釋性和透明度。

This problem plagues tons of potentially useful predictive models,

這個問題困擾著非常多 可能很有用的預測模型，

such as those that could be used to predict which bank customers

比如用來預測哪些銀行客戶最有可能

are most likely to repay a loan,

會償還貸款的模型；

or which prisoners are most likely to reoffend if granted parole.

或者預測哪些囚犯獲得假釋後 最有可能再犯的模型。

Without knowing why AI systems reach their decisions,

若不知道人工智慧系統 做出決定的原因，

many worry we can’t think critically about how to follow their advice.

許多人會擔心我們無法以批判性的 方式去思考如何遵循它們的建議。

But the transparency problem doesn’t just prevent us

但透明度問題不僅 讓我們無法了解這些模型，

from understanding these models,

it also impacts the user’s accountability.

還會影響使用者的責任歸屬。

For example, if the AI's prediction led you to break up with Alex,

比如，如果人工智慧的預測 導致你和艾力克斯分手，

what explanation could you reasonably offer them?

你能提供什麼合理的解釋？

That you want to end your happy relationship

你能說你想結束你幸福的關係

because some mysterious machine predicted its demise?

是因為某個神秘的機器預測 這段關係沒有未來嗎？

That hardly seems fair to Alex.

那對艾力克斯來說不公平吧？

We don’t always owe people an explanation for our actions,

我們不見得一定得要 向別人解釋我們的行為，

but when we do,

但當需要解釋時，

AI’s lack of transparency can create ethically challenging situations.

人工智慧透明度不足就可能 會讓情況變成有倫理爭議。

And accountability is just one of the tradeoffs we make

當我們把重要的決策 外包給人工智慧時，

by outsourcing important decisions to AI.

我們要取捨的還不僅是責任歸屬。

If you’re comfortable deferring your agency to an AI model

如果你覺得可以將你的行動權 委託給人工智慧模型，

it’s likely because you’re focused on the accuracy of the prediction.

那有可能是因為你把焦點 放在預測的準確率上。

In this mindset, it doesn’t really matter why you and Alex might break up—

若是這種心態，你和艾力克斯 為什麼分手就不重要了——

simply that you likely will.

反正你們就是可能會分手。

But if you prioritize authenticity over accuracy,

但是，如果你優先考量 真實性而非正確性，

then you'll need to understand and appreciate the reasons

那你得理解並重視

造成你們未來離婚的理由， 然後才能在今天斷絕關係。

for your future divorce before ending things today.

Authentic decision making like this is essential for maintaining accountability,

如果要確保責任有所歸屬， 像這樣的真實性決策就很重要，

and it might be your best chance to prove the prediction wrong.

且這可能是你證明 預測錯誤的最佳機會。

On the other hand, it’s also possible the model

另一方面，模型也可能

already accounted for your attempts to defy it,

已經將你試圖違抗 預測的嘗試納入考量，

and you’re just setting yourself up for failure.

你只是讓你自己走上失敗之路。

95% accuracy is high, but it’s not perfect—

95％ 的正確率是很高， 但不是完美的，

that figure means 1 in 20 couples will receive a false prediction.

它意味著每二十對夫妻中 就有一對會得到錯誤的預測。

And as more people use this service,

隨著越來越多人使用這項服務，

the likelihood increases that someone who was predicted to divorce

使用者更有可能單單因為 被人工智慧預測會離婚

will do so just because the AI predicted they would.

就決定分手。

If that happens to enough newlyweds,

如果有夠多的新婚夫妻 發生這樣的狀況，

the AI's success rate could be artificially maintained

人工智慧的成功率 就可以用人為的方式維持，

or even increased by these self-fulfilling predictions.

或甚至因為自我實現的預測而提高。

Of course, no matter what the AI might tell you,

當然，不論人工智慧告訴你什麼，

whether you even ask for its prediction is still up to you.

要不要徵詢人工智慧的預測 還是取決於你自己。