評論
打印
I realized that, as I worked on improving AI task by task, dataset by dataset,
我了解到，當我努力在改善人工智能，一個任務一個任務、一個數據集一個數據集地改善，
that I was creating massive gaps, holes and blind spots in what it could understand.
結果我卻在它能了解什么上創造出了大量的隔閡、漏洞以及盲點。
And while doing so, I was encoding all kinds of biases.
這么做的時候，我是在把各種偏見做編碼。
Biases that reflect a limited viewpoint, limited to a single dataset
這些偏見反映出受限的觀點，受限于單一數據集，
biases that can reflect human biases found in the data, such as prejudice and stereotyping.
這些偏見能反應出在數據中的人類偏見，比如偏袒以及刻板印象。
I thought back to the evolution of the technology that brought me to where I was that day
我回頭去想一路帶我走到那個時點的科技演化，
how the first color images were calibrated against a white woman's skin,
第一批彩色影像如何根據一個白種女子的皮膚來做校準，
meaning that color photography was biased against black faces.
這表示，彩色照片對于黑皮膚臉孔是有偏見的。
And that same bias, that same blind spot continued well into the '90s.
同樣的偏見，同樣的盲點，持續涌入了20世紀90年代。
And the same blind spot continues even today in how well we can recognize different people's faces in facial recognition technology.
而同樣的盲點甚至持續到現今，出現在我們對于不同人的臉部辨識能力中，在人臉辨識技術中。
I thought about the state of the art in research today, where we tend to limit our thinking to one dataset and one problem.
我思考了現今在研究上發展水平，我們傾向會把我們的思路限制在一個數據集或一個問題上。
And that in doing so, we were creating more blind spots and biases that the AI could further amplify.
這么做時，我們就會創造出更多盲點和偏見， 它們可能會被人工智能給放大。
