The New Readers of Text
Jiang Wu · 2026-04-08
On April 7, 2026, Tencent News launched a CLI tool. It is a command-line interface, with no app redesign or feed optimization. In a terminal you can look up live trending topics, pull morning and evening briefings, search AI-curated content by topic, and connect to WeCom and DingTalk for automated delivery. A news product built a reading entry point for machines. Tencent News, as a complete source of information, is reaching out to the audience that suits its content better and pushing content to it. That audience may not be a person. People read less text every year The data is clear. Video accounts for more than 75% of mobile cellular data, over 111 EB a month. On social platforms, video gets 12 times as many shares as text-and-image posts. YouTube Shorts ad impressions grew 150% year over year, and the vertical short video has become the default container for attention. Attention itself is changing too. A 2026 report cited as from Nielsen Norman Group puts average screen attention at 43 seconds, down from 47 two years earlier. A longitudinal study reaches a more extreme figure: average attention has fallen to 7.6 seconds, down 36.7% from 2000. It shows up in reading behavior. A tracking study cited from the American Psychological Association (2014 to 2024) found that adults' time spent on deep reading fell 39%, and standardized reading-comprehension scores for 18-to-34-year-olds fell 17%. A 2024 experiment at the University of Science and Technology of China was more direct: the short-video group's sustained reading attention dropped 31%, and their scores on complex text comprehension were 22% lower. Images and video are taking over attention's home ground. From the viewpoint of user behavior, text, at least long text, keeps losing importance as a form of information consumption. Text is more efficient for AI to read
This is a paid article. Subscribe to read the full text.
Home · All articles · Topics · 简体中文