三成人分不清AI假臉
We are living in the era of “deepfakes” — deep learning technology can fabricate videos with celebrity faces spliced onto porn actor bodies, or put words in the mouths of politicians or anyone else.
我們身處在「深假」（Deepfake）的年代，深度學習科技可以製作造假影片，將明星的面孔拼接在色情片演員的身體上，或讓政客或其他人說出他們從未說過的話。
deepfake(n.)深假deep learning 深度學習fabricate(v.)捏造、虛構、偽造=make upsplice(v.)拼接=attach
To help people navigate through the ever more amount of misinformation and deception online, University of Washington professors Jevin West and Carl Bergstrom created a website called WhichFaceIsReal.com.
為了幫助人們辨識網路上越來越多的假訊息和詐騙手段，華盛頓大學教授威斯特和格伯斯川創建了一個名為「WhichFaceIsReal.com」的網站，
navigate(v.)(常指借助地圖)導航、確定…的方向、瀏覽;訪問(網站)misinformation(n.)錯誤的資訊、假消息=false informationdeception(n.)欺騙deceive(v.)欺騙
The site features an online game that challenges players to distinguish AI-generated fake faces from real human ones.
網站有個線上遊戲，挑戰玩家是否能辨別AI生成的假臉與真人的臉孔。
distinguish A from B分辨兩者的差別=differentiate A from B=tell A and B apart=discriminate between A and B=tell the difference between A and B比較discriminate against~歧視
The lifelike faces were generated using an AI algorithm called StyleGAN, which stands for style generative general adversarial network.
遊戲使用StyleGAN人工智慧演算法，生成逼真的假臉，
stand for~代表be short for~是…的縮寫
Developed by US graphics processor company Nvidia, the algorithm has two neural networks “play a game of cat and mouse.”
這項技術由美國圖形處理器公司輝達開發，由兩個人工神經網路「玩貓捉老鼠的遊戲」，
Developed by...=While the algorithm is developed by…artificial neural networks 人工神經網路
One network learns from a database of real faces and creates an artificial image, while the other helps it improve by guessing if the face is real or not.
一個網路從真實人臉的數據庫中學習並創建人工智慧圖像，另一個網路則判斷該面孔是真臉還是假臉，協助優化圖像。
If it can tell the difference, the generator is sent back to the drawing board to improve its work.
若能夠分辨出差異，則生成圖像的網路就必須重新產生新的臉譜。
the drawing board 畫架
Trained for a week on an enormous data set of portraits, the image-creating network became capable of producing striking images of nonexistent people.
經過為期一周大量的肖像數據訓練，創建圖像的網路開始能生成逼真程度驚人的圖像。
Trained for a week…=After the image-creating network is trained for a week…portrait(n.)肖像畫be capable of~能夠(勝任)striking(adj.)異乎尋常的、驚人的、引人注目的
The same technique has been used to forge identities and create so-called deepfake videos.
同樣的技術也曾被用來偽造身份，以及創造所謂的「深假」（Deepfake）視頻。
forge(v.)假造;仿冒=fakea forged passport造假的護照a forged signature偽造的簽名
West and Bergstrom said they want to spread the word that the technique to generate believable, lifelike faces is now possible and proliferating online.
威斯特和格伯斯川表示，他們希望讓大家知道，目前有這樣的技術，能夠產生可信逼真的假臉孔，且網路上開始出現大量的假臉照片。
believable(adj.)可信的unbelievable(adj.)令人無法置信的proliferate(v.)激增=increase rapidly
Half a million people have visited their website and around 70 percent of the players choose correctly when trying to discern fake from real after training.
目前約有五十萬人造訪了他們的網站，大約七成的玩家在訓練後可成功辨別真假人臉。
discern(v.)看出;辨別出;明白=perceive=make out=recognize
There are a number of tells, including asymmetrical faces, misaligned teeth, unrealistic hair, as well as inconsistencies in the background of the photo.
辨別照片真偽的關鍵包括不對稱的面孔、牙齒未對準、頭髮看起來太假和照片背景中的不一致。
tell(n.)(撲克牌用語)馬腳symmetrical(adj.)對稱的asymmetrical(adj.)不對稱的misaligned(adj.)沒有對準的aligned(adj.)對準的inconsistency(n.)不一致
“We want to bring public awareness to this technology,” said West. “Just like when people started to realize that you could Photoshop images, we want the public to know that AI can replicate human faces.”
威斯特說，我們想讓大眾了解到有這項技術存在，就像人們開始意識到可以使用Photoshop修圖一樣，我們希望大眾知道AI可以複製人臉。
raise public awareness of~喚起大眾對於…的意識
“That will hopefully make people start to question things they see in different ways,” he said.
他說，這有望使人們開始意識到眼見不一定為憑。