The barriers to talking about big models: who knows better about users, who deserves more

NOVEMBER 9TH, ACCORDING TO THE FIRST FINANCIAL REPORT, DURING A DIALOGUE AT THE 2026 SCIENCE PRIZE FOR THE FUTURE, ZHANG YAQIN ASKED THE CHIEF SCIENTIST AIYao Sun YuHow does a model company retain users when the developers can switch to other models and even combine several intelligent bodies

6323f930j00to7mlf008vd000hs00a0m

Yao Sun-yu believes that more persistent barriers stem from unique data available to the product, an understanding of the user situation, and how much the user is willing to give it。

HE GAVE THE EXAMPLE OF A TWEET. IF A USER CAN USE AI DIRECTLY IN A MICRO-MAIL, CALENDARS, CONTACTS AND ACTIVITY INFORMATION CAN HELP IT UNDERSTAND THE USER ' S NEEDS. EVEN IF ANOTHER MODEL IS MORE CAPABLE, IT IS DIFFICULT TO PROVIDE THE SAME SERVICE WITHOUT SUCH INFORMATION。

YAO SUN YU ALSO MENTIONED THAT THE WILLINGNESS OF USERS TO OPEN THEIR PERSONAL INFORMATION ONCE AI HAD BEGUN TO PERFORM ITS TASK FOR USERS DEPENDED ON THE USER ' S TRUST IN THE PROCESSING OF THE DATA AND THE TASK。

Turning to the open source of models, he cited the example of cybersecurity, noting that, with increased generic reasoning and programming capacity, models could learn to exploit loopholes even if they did not specifically learn about attack content. He advocated automating and manual assessment prior to publication and continuing to monitor harmful calls after they became available。

statement:The content of the source of public various media platforms, if the inclusion of the content violates your rights and interests, please contact the mailbox, this site will be the first time to deal with.
Information

Rotnus is asking to speed up AI research and development and to promote Siri's initiative in the task

2026-11-9 8:54:44

Information

QUESTION AND ANSWER AND GENERAL SEARCH

2026-11-9 8:58:27

Zenium
Search