Page 228 - AI for Good-Innovate for Impact

AI for Good-Innovate for Impact

• SDG 4 Ensure inclusive and equitable quality education and promote lifelong learning
opportunities for all: Providing accessibility products and services for people with
disabilities has addressed their learning and communication barriers, enabling them
to access education and learning resources on a more equal basis and promoting
educational equity.
• SDG 8 Promote sustained, inclusive and sustainable economic growth, full and
productive employment and decent work for all: Through the implementation of the
accessibility project, vivo demonstrates a strong sense of social responsibility while also
promoting the development of the accessibility technology and market expansion,
providing more job opportunities for people with disabilities and facilitating their
integration into society.

53�2�2 Future Work:

1. Data Collection: For the usage scenarios of individuals with disabilities, we aim to expand
the training datasets to better meet user needs and develop more targeted solutions.
This will further improve model performance.
2. Model Development: We plan to upgrade and iterate the capabilities of our current
multimodal large models to enhance their performance, reduce latency, and add real-
time conversational abilities, thereby improving the user experience and continuously
introducing innovative features. Additionally, we seek collaboration with government
agencies, social organizations, enterprises, and developers to build consensus and
promote the application of accessible technology, fostering the construction and growth
of information accessibility.

The development of vivo's accessibility project is a long-term endeavor that requires ongoing
refinement and upgrades. Our goal is to deliver better services and products to individuals with
disabilities. To achieve the building and development of information accessibility, we need to
support and enhance the innovation and application of accessible technologies. Accessible
technology should serve as a bridge that enables more individuals with disabilities to integrate
into society, enjoy a fulfilling life, and realize their personal value.

53�3� Use Case Requirements:

• REQ-01: Algorithm Capabilities: The multimodal large model uses LLM as its base
and adds a visual module to LLM, enabling the model to simultaneously process data
from both text and image modalities, which can better handle complex and real-world
problems.
• REQ-02: Hardware Requirements: Users are required to use vivo smartphones that
support accessibility features to ensure that the devices can run high-performance
algorithms and handle complex tasks.
• REQ-03: User Feedback Collection and Analysis Technology: By collecting user feedback
on the accessibility feature via online and offline channels, we use data analysis technology
to evaluate these feedbacks and optimize the accessibility feature accordingly. Such
an analytical method aids in understanding user needs and behaviors, thus enhancing
feature design and boosting user satisfaction.
• REQ-04: Social Development Trends and Corporate Responsibility: Given the aging
population and the growing demand for accessibility in social, vivo has a responsibility to
provide accessible products. This enables AI to permeate all aspects of life through mobile
phones, supporting the quality of life for users with visual and hearing impairments.

In order to meet the above use case requirements, vivo's accessibility project combines
technology, hardware, requirements, and corporate responsibility to provide high-quality

212