How AI Personas Revitalize Baidu App Comment Sections: Design Insights & Results
This article examines Baidu's design experiments that embed AI characters into the app's comment area, detailing user research, evolution of comment communities, validation of AI assistants and film‑related roles, persona creation, and future plans that together boost engagement and earn international design awards.
Introduction
With the rapid development of artificial intelligence, designers are exploring ways to embed AI naturally into user‑facing scenarios. This article shares Baidu APP’s experiments with AI characters in the comment area, from design insights to validation and persona creation.
Design Insight
Analysis of user research and business data showed that negative feelings in comment sections stem from a cold atmosphere and lack of replies, reducing users’ willingness to comment. The team asked how to integrate AIGC capabilities to make commenting more attractive and interactive.
Large language models can generate sentences with specific tones via prompts. Existing products place a virtual person on a separate page, which creates a deep user path and unfamiliarity. Baidu chose a native‑style design that embeds AI directly into the comment flow.
Evolution of Comment Communities
First generation: Early internet comments served mainly as feedback, with little interaction.
Second generation: Social media added reply functions, enabling broader interaction.
Third generation: Blogs allowed authors to reply, fostering tighter creator‑user connections.
The next step is a “new world” where AI characters join user‑to‑user and user‑to‑author interactions, forming a network of “user + author + multiple AI roles”.
AI Role Validation
To test the concept, a “video AI assistant” was placed in the comment area of hot‑topic videos. Prompt‑driven responses summarized video knowledge and added depth, increasing comment value.
Data showed the AI assistant significantly boosted user comments, and interaction willingness with AI roles was dozens of times higher than with real comments.
Subsequently, over 60 AI film‑related characters were launched, each with distinct personalities, catchphrases, and tones. Interaction metrics for some AI characters even surpassed the assistant.
Analysis of user replies revealed distinct user groups—passionate, detective, opposing, neutral, and spiritual—each reacting differently to AI characters, which further enlivened the discussion.
For “spiritual” users, a proactive reply feature was added, and top AI characters gained private‑chat capabilities, allowing 1‑to‑1 conversations.
Persona Creation
Virtual Assistant
Three visual concepts were explored: realistic peers, futuristic tech personas, and hybrid bio‑digital humans. User testing favored clean, friendly female avatars, leading to a bio‑digital human direction.
Midjourney was used to generate three personality styles, and the final avatar was selected.
Film‑Related Characters
Avatars combined iconic lines, scenes, and memes, often using dynamic facial expressions to increase “personality feel”. Background images on personal pages incorporated catchphrases, enhancing realism.
Future Outlook
Beyond current features, Baidu plans to expand AI roles to act like real KOLs—publishing videos, posts, interacting, and maintaining fan relationships, creating a closed‑loop AI matrix.
The design has earned multiple international awards, including Italy’s A’Design Gold, USA’s MUSE Gold, London Design Platinum, and France Design awards.
Designers are encouraged to break conventions, hypothesize boldly, and let AI deliver real value in practical scenarios.
Baidu MEUX
MEUX, Baidu Mobile Ecosystem UX Design Center, handling end-to-end experience design for user and commercial products in Baidu's mobile ecosystem. Send resumes to [email protected]
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.
