Industry

Microsoft AI, Alibaba AI Outscore Humans in Reading Comprehension

Jan 17, 2018, 11:20am2020/07/21 14:33:52 Sina Tech

AI has defeated humans in chess and poker, and now it has come to dominate another field: reading comprehension.

SEE ALSO: Alibaba Develops Autonomous Driving: Champions of Three Tasks

The AI developed by Microsoft and Alibaba<img decoding="async" class="trade-card__loading" layout="fixed" height="50" width="50" src="https://assets.pandaily.com/uploads/2021/12/loading.png" placeholder /> beat humans in a recent reading test. The data came from Stanford University. Each AI analyzed more than 500 articles on Wikipedia and matched questions and answers. In total, Standford prepared more than 100,000 pairs of data sets for testing.

One sample question was “when did Genghis Khan die?” The answer was AD 1227.

“Machines used to be unable to surpass humans in this test. This was the first time,” Alibaba<img decoding="async" class="trade-card__loading" layout="fixed" height="50" width="50" src="https://assets.pandaily.com/uploads/2021/12/loading.png" placeholder /> said in a statement.

The Microsoft program scored 82.6 points and Alibaba<img decoding="async" class="trade-card__loading" layout="fixed" height="50" width="50" src="https://assets.pandaily.com/uploads/2021/12/loading.png" placeholder /> scored 82.4 points, both surpassing the average human score of 82.3. Other AI programs have were also tested and confirmed to surpass human scores, such as the AI developed by the Allen AI Institute, Tencent<img decoding="async" class="trade-card__loading" layout="fixed" height="50" width="50" src="https://assets.pandaily.com/uploads/2021/12/loading.png" placeholder />, Salesforce and others.

This article originally appeared in Sina Technology and was translated by Pandaily.