Let AI "Speak People": Microsoft Copilot online test Real Talk, sincerely speaking to you

News of January 20, Windows Latest published yesterday, January 19th, reporting thatMicrosoftIn response to the fierce competition between Google Gemini and ChatGPT, the government has decided to take a decision on the issueInviting global user testing Copilot , and start testing video generation。

Let AI "Speak People": Microsoft Copilot online test Real Talk, sincerely speaking to you

Microsoft is currently facing serious market challenges, and SimilarWeb, a network analyst, shows that the market share of the Copilot page is only around 1%. In order to reverse the situation and counterbalance with Gemini and ChatGPT, Microsoft has not slowed down, and two core features, Real Talk and Video Generation, were recently introduced。

First of all, Real Talk mode. Unlike Sydney, who was out of control in the early days, the core is to adjust AI's response by two dimensions: "Depth" and "Wringing style"。

When the user initiates the dialogue, the Copilot will first draw on past memories and automatically select the appropriate depth (e.g. standards, compression) and style (e.g. leisure). Microsoft stressed that the model allowed users to look at Copilot's thinking processes at all times, with the aim of transforming AI from a simple assistant to a private friend who could engage in “class conversations”。

In practical experience, Real Talk shows a very different character. Instead of responding carefully to questions in a “corporate disguise”, it would show real curiosity。

When users raise unreasonable questions, they even counter and share their true interest in the subject, instead of pretending to be enthusiastic about every issue. By providing alternative perspectives and more stimulating interactions, such mechanisms have significantly increased the vitality of the dialogue。

In the area of visual capability, the media found a new "Create a video" switch in the Andreau Copilot version. Test shows that this feature allows the user to generate video clips of up to 8 seconds with audio。

It is not clear whether the function is based on Microsoft Self-Research Model or OpenAI Sora Model, but it is noteworthy that it is available under the basic Microsoft 365 subscription and does not appear to have a payment threshold。

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.
HeadlinesInformation

MEN DELIBERATELY USED AI TO PRODUCE LOW-LIFE VIDEOS AT MULTIPLE LANDMARKS AND WERE HELD IN ADMINISTRATIVE DETENTION

2026-1-20 11:44:46

HeadlinesInformation

HU YUN RELEASED 2025 CHINA AI ENTERPRISE 50 STRONG, WITH COLD WU QI VALUED AT 63 BILLION YUAN

2026-1-20 11:48:22

Search