Page 319 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

them directly to service or sales teams. For example, in customer service, our "AI-assisted +
human-in-the-loop" model uses ASR for real-time transcription, natural language processing
for intent extraction, and retrieval-augmented generation for instant answers. AIGC generates
summaries, together improving service efficiency by 24% and boosting marketing by 15%. 4.3 - 5G
The 24/7 chatbot reduces night shifts by 30%, improving working conditions for employees.
In terms of "Economic Growth,": CloudTop leverages Transformer-based models for deep
intent understanding and tailored solutions, enabling efficient, low-cost digital transformation.
This helps large firms improve productivity and SMEs accelerate innovation, driving high-quality
digital growth. Through cluster analysis of customers' internal behavioral information and
external characteristic information in the enterprise, different risk scenarios are preset according
to the usage of security products. When customers initiate consultations, we will accurately
identify whether they need security protection by combining customer intent, clustering results,
and risk scenarios. At the same time, customized security reinforcement and highly available
security solutions are provided based on customers' resource usage to support the sustainable
development of customer businesses and promote healthy economic growth.

Its standardized APIs and container-based deployment make integration quick and easy,
allowing all industries to scale AI-powered marketing and services, enhance staff productivity,
and increase business value.

2�3 Future Work

Future work on this case will focus on model development, creating new variations/extensions
for the same use case, and standards development related to the use case. We plan to expand
our case's influence in two key areas, application optimization and social contribution.

Application optimization:

The resources from the awards will be allocated to support the ongoing development, platform
construction, and promotion of the CloudTop. Expand the application scope of Agents from
post-event knowledge extraction and quality inspection to mid-event and pre-event stages,
establish the capability to predict customer behavior, provide prompt capabilities for correct
behaviors and interception capabilities for abnormal behaviors for service and marketing
personnel, and eliminate penalties for employee errors. We will do further research on semantic
understanding, sentiment analysis, and multi-round dialogue technologies to enhance the
interactive experience and problem-solving efficiency of robots. We will also optimize machine
learning algorithms to improve the accuracy and response speed of intelligent agents, and
enhance reasoning capabilities. Additionally, the dataset will be improved continuously, and
a large-scale high-quality corpus and a high-level data annotation mechanism will be built
to promote model iteration and upgrading. Based on the high-quality customer service and
marketing dataset precipitated from deep applications, Agents refine the dataset and open
it to the public, promoting collaborative innovation across industries.Meanwhile, we will keep
a close eye on regulatory requirements and the forefront of industrial technologies, and
continuously strengthen data security and privacy protection.

Social contribution:

The resources will be utilized to strengthen cooperation with the International Telecommunication
Union (ITU) and contribute to the formulation of relevant ITU standards. Based on the deep

283