Artificial intelligence-powered computers have already beaten humans at board games like Chess and Go, they’re now capable of beating humans at reading comprehension tests as well. Alibaba announced today that its AI was able to beat humans in a global reading comprehension test. The AI has been developed by Alibaba’s research unit, the Institute of Data Science of Technologies.
Alibaba’s AI attained a score of 82.44 in Exact Match on the Stanford Question Answering Dataset, narrowly beating up the human score of 82.304. The assessment of the reading comprehension was conducted by ranking the computer’s answers and comparing them with average human responses. Microsoft and Alibaba were able to win this by the slimmest of margins.
The Stanford Question Answering Dataset has more than 100,000 question and answers sets that are based on over 500 Wikipedia articles. Alibaba explained that its AI was able to win because its neural network model is based on the Hierarchical Attention Network which enables the AI to read from “paragraphs to sentences to words” in order to identify phases that can have potential answers.
“We believe the technology underneath can be gradually applied to numerous applications such as customer service, museum tutorials, and online responses to medical inquiries from patients, decreasing the need for human input in an unprecedented way,” said the chief scientist at Alibaba’s research arm.
China’s government is also a big backer of AI development and the country’s native tech giants like Alibaba, Baidu, and Tencent are all heavily invested in AI efforts to leap ahead of the West.