谷歌識(shí)圖去標(biāo)簽：因性別標(biāo)簽可能會(huì)制造或增加偏見(jiàn)

Google has announced that its image recognition AI will no longer identify people in images as a man or a woman. The change was revealed in an email to developers who use the company's Cloud Vision API that makes it easy for apps and services to identify objects in images.

谷歌宣布，其圖像識(shí)別人工智能將不再識(shí)別圖像中人物的性別。在發(fā)給使用谷歌云視覺(jué)API的開(kāi)發(fā)者的一封郵件中，谷歌透露了這一改變，谷歌的API使應(yīng)用程序和服務(wù)可以更容易地識(shí)別圖像中的物體。

In the email, Google said it wasn't possible to detect a person's true gender based simply on the clothes they were wearing. But Google also said that they were dropping gender labels for another reason: they could create or reinforce biases. From the email:

谷歌在郵件中表示，只通過(guò)一個(gè)人的衣著是無(wú)法判斷其真實(shí)性別的。但谷歌也表示，他們摘掉性別標(biāo)簽還有另一個(gè)原因：那就是性別標(biāo)簽可能會(huì)制造或增加偏見(jiàn)。郵件中寫(xiě)道：

Given that a person's gender cannot be inferred by appearance, we have decided to remove these labels in order to align with the Artificial Intelligence Principles at Google, specifically Principle #2: Avoid creating or reinforcing unfair bias.

鑒于通過(guò)外表并不能推斷出一個(gè)人的性別，我們決定去掉這些標(biāo)簽，以與谷歌的人工智能準(zhǔn)則保持一致，特別是第二項(xiàng)原則：避免制造或增加不公平的偏見(jiàn)。

As for what Google's second AI principle states, it says:

谷歌的第二項(xiàng)人工智能準(zhǔn)則指出：

AI algorithms and datasets can reflect, reinforce, or reduce unfair biases. We recognize that distinguishing fair from unfair biases is not always simple, and differs across cultures and societies. We will seek to avoid unjust impacts on people, particularly those related to sensitive characteristics such as race, ethnicity, gender, nationality, income, sexual orientation, ability, and political or religious belief.

人工智能算法和數(shù)據(jù)集可以反映、增加或減少不公平的偏見(jiàn)。我們意識(shí)到，區(qū)分公平和不公平的偏見(jiàn)并不總是容易的，而且在不同的文化和社會(huì)中也會(huì)有所不同。我們會(huì)努力避免對(duì)人們產(chǎn)生不公平的影響，特別是那些與敏感特征有關(guān)的影響，如種族、民族、性別、國(guó)籍、收入、性取向、能力以及政治或宗教信仰。

Instead of labeling someone a man or woman in an image, that individual will now just be classified as a "person" by Google's Cloud Vision API image recognition tools. The move should be a welcome step in the right direction for those who have long pointed out that Google and other tech companies' systems have an issue correctly categorizing people who identify as trans. Per Business Insider's testing, it appears that the labeling changes have already taken place, so developers who use the API should see the changes immediately.

如今，在谷歌的云視覺(jué)API圖像識(shí)別工具中，人們?cè)趫D像中不會(huì)被貼上性別標(biāo)簽，而是被歸類(lèi)為“人”。長(zhǎng)期以來(lái)一直有人指出，谷歌和其他科技公司的系統(tǒng)在正確判定變性人方面存在問(wèn)題。對(duì)他們而言，谷歌的這一舉動(dòng)可以說(shuō)是朝正確的方向邁出的可喜的一步。根據(jù)Business Insider的測(cè)試，谷歌的標(biāo)簽似乎已經(jīng)產(chǎn)生了變化，因此使用API的開(kāi)發(fā)人員應(yīng)該也能即刻看到這種改變。