請各位先登陸 (studentID,Moodle 密碼登陸),後編輯
請各位使用左方之工具列上的上載文件連結上載習作相關的相片

Journalists wins A.I. over context recognition

出自香港新聞網 - 樹仁新傳系學生實習習作

跳轉到: 導航, 搜尋

Structure: 8 Content: 7.5

By Matthew Chan November 23, 2018


Artificial intelligence (A.I.) is not good at recognizing context, an expert says.

Bosco Yung, machine learning engineer said it is still difficult for immediate reporting the emergency situations such as terrorist attack by A.I., despite the flourishing A.I. development recently.

He said news that are more routine in terms of format, such as figures and studies reporting, it is likely that A.I. robot could write a ‘nice’ news article.

“In the big data era, A.I. could do a lot of analysis, collect a lot of data, in the end A.I. could also give suggestions and predictions, but human being is still the final guard,” said Yung.

Know as ‘automated journalism’, A.I. technologies have been adopted in several well-known newsrooms such as New York Times, Reuters, as well as the world’s first A.I. TV anchor unveiled by Xinhua News Agency in China.

Yung said A.I. presently select news from the popularity of topics, include comparing the search rate of different keywords.

However, “human is worrying and need to do fact check as the accuracy is not hundred percent,” said Yung.

He said the accuracy in distinguishing real and fake news could be improved by the advanced A.I. in the future, through consideration of interrelationships between information.

A.I. should be trained to ‘dig into the source’ and to judge ‘whether the sources are reliable’ by tracing the source accounts and their followed accounts, said Yung.

“Machines have to judge whether celebrities are actually giving opinions on topics they know,” said Yung.

He said journalist therefore should learn to use A.I. tools carefully and do fact check frequently.

個人工具