Prof. Qiang Yang, founding head of Huawei Technologies' AI research lab, believes that big data paves the way for greater use of artificial intelligence in finance. Extracts from a recent interview.What’s the next step for AI?Artificial Intelligence has been successful mostly in fairly limited and well understood domains. A prime example is Go, where the rules are set and the range of data is confined to predetermined board locations. In the area of chatbots, the best chatbot today can do well in vertical, targeted domains, but cannot yet solve general problems in conversations that cut across many different areas.In the future, we will see more AI successes where high-quality data are available and rules are increasingly well-understood in terms of possible operations and data. These will include areas of finance, online education, manufacturing and healthcare.Can AI be used to forecast financial markets?AI’s uses have dramatically increased with the availability of big data. Machine-learning models can be trained to help make forecasts of future market movements by using information on all market transactions and movements over several years.AI can also excel in text analysis of mountains of news, documents, financial reports and company annual reports to detect important market signals – including from social media.
Topics:
Perspectives Community Manager considers the following as important: AI, Artificial intelligence, data mining, financial market forecasting, Macroview
This could be interesting, too:
Claudio Grass writes 2023: A year in review
Cesar Perez Ruiz writes Weekly View – Big Splits
Cesar Perez Ruiz writes Weekly View – Central Bank Halloween
Cesar Perez Ruiz writes Weekly View – Widening bottlenecks
Prof. Qiang Yang, founding head of Huawei Technologies' AI research lab, believes that big data paves the way for greater use of artificial intelligence in finance. Extracts from a recent interview.
What’s the next step for AI?
Artificial Intelligence has been successful mostly in fairly limited and well understood domains. A prime example is Go, where the rules are set and the range of data is confined to predetermined board locations. In the area of chatbots, the best chatbot today can do well in vertical, targeted domains, but cannot yet solve general problems in conversations that cut across many different areas.
In the future, we will see more AI successes where high-quality data are available and rules are increasingly well-understood in terms of possible operations and data. These will include areas of finance, online education, manufacturing and healthcare.
Can AI be used to forecast financial markets?
AI’s uses have dramatically increased with the availability of big data. Machine-learning models can be trained to help make forecasts of future market movements by using information on all market transactions and movements over several years.
AI can also excel in text analysis of mountains of news, documents, financial reports and company annual reports to detect important market signals – including from social media. For example, if a company’s annual report implies that several new products will be launched and social media indicate that similar products have long been in demand, machine reading could predict that the products will sell very well.
Will AI be able to predict financial market prices within the next few years?
I believe that with sufficient data, AI algorithms are in place to make more accurate predictions of market movement than an individual human analyst. In financial markets with enough historical data, it is possible to assess general market situations and make more than random assessments of market actions.
However, it is often difficult to accumulate a sufficient amount of the complete data needed to create an AI system that could surpass human performance. The data can be extremely sparse or noisy, and it may be difficult to understand fully the meaning of various fields in it. To address these challenges, researchers are investigating ways to combine expert rules with statistical learning from the data.