Page 748 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

Objectives and Aims

This project leverages on-device AI technology to:

• Develop real-time, secure, and efficient assistive tools addressing core needs of individuals
with hearing/visual impairments and multilingual groups (e.g., voice interaction, sign
language translation, offline support).
• Break Technical Barriers: Reduce cloud dependency through localized AI processing,
ensuring privacy and energy efficiency, and advancing the "AI for Good" vision.
• Promote Social Equity: Bridge the digital divide for disabled communities, enhancing
social participation and life autonomy.

Core Technological Breakthroughs:
• On-Device Lightweight AI Models: Integrate deep learning with edge computing to
optimize efficiency across hardware configurations (e.g., vivo Voice, vivo Sight).
• Multimodal Fusion: Combine voice, image, and tactile perception (e.g., sign language
recognition, screen reading) for comprehensive accessibility.
• Offline-First Design: Enable critical functions (e.g., Sound Recognition, Offline Audio
Recognition) without network connectivity.
Key Features:

• Voice/Accessibility Calls: Real-time speech-to-text and multilingual translation for
seamless communication.
• Sign Language Translator: China’s first AI-powered tool converting text to continuous
sign language via virtual avatars. Sign language experts are engaged throughout the
product lifecycle. During data collection, they assess data accuracy and practicality, and
also help validate standards, protocols, and data usability. In test/launch phase, they
provide recommendations and product experience reports. vivo Sight: Visual multimodal
technology providing environmental descriptions and image Q&A for visually impaired
users.

Social Impact:

• Reduced Inequality: Serve over 500 million global users (300 million in China),
empowering 80 million individuals with disabilities and 300 million elderly users through
digital literacy and employment skills.
• Sustainable Production: On-device AI reduces cloud energy consumption by 90%,
advancing green computing and privacy protection.

Industry Influence:

• Standardization: Collaborate with governments and Non-Governmental
Organizations(NGOs) to establish accessibility technology standards.
• Ecosystem Collaboration: Launch an open developer platform to foster third-party
innovations.
• Technological Evolution: Continuously upgrade multimodal models for real-time
interaction and predictive accuracy, building an inclusive "Information Accessibility"
ecosystem.

Impact on healthcare:

Reduced Inequality

By providing AI-assisted features on the device, the use case provides more equal opportunities
for access to technological services for disable groups (such as people with visual or hearing

712