In recent years, Artificial Intelligence has made significant advancements with the emergence of large language models like OpenAI’s GPT. These large language models possess new capabilities to assist humans, such as answering questions, summarizing text, generating content, multilingual translation, and programming support. However, as inherently generative AIs, they generate natural language based on probability predictions from trained data, and thus, these large language models currently lack the logical reasoning ability to explain real-world business problems. Building a foundational model with general inference capabilities will require extensive breakthrough research in core technology, a direction that the global scientific community is currently striving towards.

In the coming years, we can aim to develop AI inference technology for specific problem domains. For example, providing legal advice to citizens to help them assess their likelihood of winning a lawsuit, or solving the problem of detecting overlaps between legal documents to support public officials.

There have been some recent studies on AI inference problems, combining large language models with knowledge graphs or incorporating a chain of thought. However, these studies are still in their early stages and have not yet met the requirements of practical applications.

We hope that the research community and technology companies will join hands to accelerate the development of AI inference models for application in Vietnam, benefiting government authorities, businesses, and citizens.