Table of Contents
Game producer Tamaki Jun unveiled the “GALLOPIA” horse racing prediction system he developed at an event called “Generating AI Can Do Anything Exhibition”. The group chat action system used several virtual AI female characters to predict the outcome of horse races through simulated conversations, and eventually won a lottery ticket with odds as high as 2275.3 times.
▲ Aya Tamaki released her self-developed horse racing prediction system “GALLOPIA”, and predicted the results of horse races through simulated conversations (Image source:gigazine)
▲ During the test, I won a trifecta of 1,000 yen and the final payout was 227,530 yen.
GALLOPIA is not a traditional horse race prediction tool, but emphasizes interactivity and fun. The system consists of 8 female AI characters with different professional fields. In the end, the “protagonist” role synthesizes all the analysis and comes up with a recommended commitment strategy. For example, pedigree analysis is managed by Chiaki, jockey and stakeholder evaluations are managed by Aster, while Hayate focuses on speed and analysis of training data. Their discussion process was analyzed by “Research”, and “Fumino” was responsible for organizing the discussion into clear recommendations.
▲GALLOPIA consists of 8 female AI characters with different professional fields (image source:gigazine)
According to Tamaki’s description, the system’s design goal is to “enable interesting human-like conversations between female AI characters to express strong opinions about a particular horse” rather than focusing on building a pre- High accuracy horse race telling.
▲ The system’s design goal is not to build a high-accuracy horse race prediction algorithm (Image source:gigazine)
Yuzhi combined several language modules during the development process and adopted “Multi-Agent LLM Orchestration” technology. The main idea is to use independent models for analysis in different professional areas and finally to integrate all the results. The models used include: gemini-1.5-pro, which is responsible for extracting advantages and disadvantages from big data, GPT-4o, which do some data analysis to help the main character make decisions, and claude -3, which creates the main arguments of each AI character. .
▲ Several language models were combined during the development process, and a “multi-agent LLM orchestra” was adopted
Calling the model through several short input propositions can avoid the problem of reduced output accuracy caused by one large input.
GALLOPIA demonstrates the ability of language models to integrate multiple areas of knowledge. Especially in the complex analysis of horse racing predictions including pedigree, jockey evaluation and speed data, combining multiple professional models to work together can greatly improve prediction accuracy. – tell Tamaki admits that the main goal of the system is to provide an interesting experience. practical AI tools.
Source of information and pictures:gigazine
The latest videos
What unique features does GALLOPIA offer compared to other AI systems in the field of sports predictions?
1. What inspired the creation of GALLOPIA, an AI system designed for horse race predictions?
2. Can you explain how GALLOPIA works and what role each of its female characters plays in predicting race outcomes?
3. How is GALLOPIA different from traditional horse race prediction tools?
4. What is the main design goal of GALLOPIA, and how does it provide an interesting experience for users?
5. Can you describe the language models used in GALLOPIA and how they integrate multiple areas of knowledge for accurate predictions?
6. How does the “Multi-Agent LLM Orchestration” technology contribute to the effectiveness of GALLOPIA’s predictions?
7. What challenges did you encounter during the development of GALLOPIA, and how did you overcome them?
8. As someone with a background in horse racing, do you think GALLOPIA could become a valuable tool for predicting races, or is it more geared towards entertainment value?
9. How might GALLOPIA evolve in the future to further improve its predictions and user experience?
10. As language models continue to advance, do you see potential for similar AI systems to be developed in other areas with complex data analysis requirements, such as sports betting or financial investments?
.jpg "Saudi Arabia’s World Cup Qualifiers Disappointment: Hervé Renard Regrets Return After Shocking Loss to Indonesia")
