AI can read and comprehend better than you

Artificial Intelligence software created by Microsoft and Alibaba squeaked by humans to take the top spots in a complex reading-comprehension test.

The Stanford Question Answering Dataset, or SQuAD, includes more than 100,000 sets of questions and answers drawn from 500 Wikipedia articles. SQuAD was created to challenge AI’s ability to understand written information.

Microsoft said its AI system scored 82.65 on the test; Alibaba’s entry reached 82.44. Both beat the human subjects’ aggregated score of 82.305. Google, Tencent, Yiwise and the Allen Institute for Artificial Intelligence were among the entries.

TRENDPOST: Artificial intelligence will become a key office assistant for professionals who must confront and digest vast amounts of written information. AI-reading technologies can cull key points in text, saving humans time that instead can be invested in delving deeper into critical details as needed.

Skip to content