“It was the user who needed to learn to use a set of software in the past, but now it is the AI that learns to serve the user,” says Antony Tseng, Head of Engineering, QA, and Operations, Datayes, in an exclusive interview with China Securities Journal, as he paints a rosy picture of the application of LLMs (Large Language Models) in the financial sector.

AI Learns to Serve

In essence, ChatGPT is the application of LLMs, which has expanded its capabilities from simple tasks (such as Q&A, text recognition and classification) to more creative ones (such as text or code generation, and human-like conversations), comments Antony. “If we apply LLMs to the processing of financial data, we can use natural language to collect, process, analyze, and make predictions with complex financial data,” adds Antony.

In Antony’s view, understanding and generating natural languages accurately and logically is one of the most obvious advantages of LLMs, which will empower financial data processing and analysis. With the development of LLMs, it will become increasingly easy to collect, process and analyze data. “There will be no need to learn to use the operating system and extract data, and with a simple command, the system will help us to complete these tasks automatically,” says Mr. Tseng.

Antony gives an example of a new feature that Datayes has recently released in the search function of its SaaS product Datayes!Pro – Search Results Integrated with LLMs. “If a user searches for a financial answer, the LLMs will generate a summarization or content of research reports and news searched based on the question, each sentence and data is not made up, but has a source, and the source will be marked, enhancing the veracity and objectivity of the text,” explains Antony.

LLMs Are Part of AI

ChatGPT has taken the Internet by storm. LLMs behind ChatGPT are part of AI applications. LLMs can enable the learning and understanding of natural languages, but making predictions based on financial data cannot rely on LLMs alone, according to Antony. “For example, Datayes uses other AI models to make predictions on the revenues of listed companies through Datayes AI Revenue Forecast, which is a combination of fundamental analysis and quantitative algorithms. We publish the resulting forecasts to the public in advance, open to market scrutiny. Comparative results over several years have shown that the accuracy has consistently exceeded consensus expectations,” says Antony.

AI Revenue Forecast is a combination of AI (Artificial Intelligence) and HI (Human Intelligence). Datayes has launched a series of products and services based on AI+HI besides AI Revenue Forecast, such as Macro Frameworks, Industry Prosperity Forecast, Knowledge Graph, and more, to explore and promote the application of AI in the financial sector.

In addition to forecasting, Antony thinks that AI is also good at the processing and analysis of alternative data, such as the collection, processing and analysis of news and public opinion. “Financial news often has a broader impact on the market sentiment, but it is very complex to analyze the positivity, negativity or neutrality of so much information. AI can qualify and quantify news and public opinion in terms of sentiment to provide investors with investment signals and leads,” says Antony.

Matrix of Fundamental + High-frequency + Alternative Data

Datayes processes and provides a variety of datasets. Its fundamental data covers all traditional financial data, with a total quantity of petabytes, which is in a dominant market position; its high-frequency data is low latency and high quality, and there are rich derivative indicators available; its alternative data leads the industry with a wide range of categories, many of which are exclusive data sources. Datayes will continue focusing on the combination of digital technologies such as AI, Big Data, Cloud Computing with professional investment concepts to further leverage AI and LLMs in search and recommendation, analysis and forecast, and decision-making support, and bring more cognitive tech based digital and intelligent products, services and experiences to investors.