Alibaba and Microsoft’s artificial intelligence (AI) programs beat humans in Stanford University’s reading comprehension test

Briefing

Alibaba and Microsoft’s artificial intelligence (AI) programs beat humans in Stanford University’s reading comprehension test

January 16, 2018

Briefing

  • Machines Beat Humans – Alibaba and Microsoft’s artificial intelligence (AI) programs scored higher than humans in Stanford University’s reading comprehension test, scoring 82.44 and 82.65 respectively, versus humans’ record of 82.304
  • Stanford Test Dataset – Large-scale reading comprehension test with more than 100,000 questions pertaining to some 500 Wikipedia articles
  • Early Testers – International universities and global technology firms, such as Google, Facebook, and IBM also used Stanford test to determine machine learning models’ reading comprehension capabilities
  • Strengths and Weaknesses – Alibaba’s model strong on questions with defined answers, but weak on questions with too vague or ungrammatical language or expressions
  • Potential Applications – Include customer service, museum tutorials, medical chatbots, targeted advertising, and social media
  • Alibaba’s AI Plans – Announced $15 billion investment in emerging technologies, including AI, in October 2017

Accelerator

Sector

Information Technology

Organization

Alibaba Group Holding Ltd., Facebook Inc., International Business Machines Corp., Microsoft Corp., Stanford University

Source

Original Publication Date

January 15, 2018

Leave a comment