AI Assistants for Video Games

Co-funding Company

Tencent

Tencent Games was established in 2003. We are a leading global platform for game development, operations and publishing, and the largest online game community in China. Tencent Games has developed and operated over 140 games. We provide cross-platform interactive entertainment experience for more than 800 million users in over 200 countries and regions around the world. Honor of Kings, PUBG MOBILE, and League of Legends, are some of our most popular titles around the world. Meanwhile, we actively promote the development of esports industry, work with global partners to build an open, collaborative and symbiotic industrial ecology, and create high-quality digital life experiences for players.

https://www.tencentgames.com/

Supervisory team

University of Edinburgh PI: Kartic Subr - k.subr@ed.ac.uk (School of Informatics)
Personal website: https://homepages.inf.ed.ac.uk/ksubr/ 

Company supervisor: Ruidong Wang - ruidwang@global.tencent.com 

Abstract

This project aims to develop a verbal (typed) chat assistant for video games equipped with generative capabilities to provide game puzzles and analyse game logs for adaptive puzzle generation. Key objectives include integrating the assistant into a selected game, chosen in collaboration with Tencent, generating game logs through gameplay or simulation, and developing a reasoning engine to predict optimal or reasonable actions. The verbalization engine will facilitate game knowledge interpretation, while a generative modelling engine will create game states for tutorials or training. Outcomes include the development of a verbal game assistant for tactical consultation [Q3], a counterfactual reasoning and analysis assistant [Q6], and a game summary generator for personalized social media sharing [Q10]. Additionally, the project will produce a customized tutorial generator addressing gameplay bottlenecks [Q14], culminating in a consolidated report and efficient codebase [Q16]. Each phase will result in research publications and new deployable features.

Project Background

Tencent UK plans to further deepen partnership with University of Edinburgh in 2026 by launching two PhD co-funded studentship, aiming to support Tencent Games’ long-term strategy in AI-driven game development, data intelligence, and next-generation interactive technologies. Through this programme, Tencent Games Data & AI team will participate in the selection and supervision of doctoral candidates, aligning academic research with real-world industry challenges and innovation needs.

Project Aims

1. Develop a verbal chat assistant (typed natural language)for video games along with its analysis.
2. Equip the assistant with generative capability to provide ‘video game puzzles’.
3. Analyse game logs to generate adaptive puzzles for customised training.
4. Bonus: Roll-out the assistant, integrated within one of the games.  

Expected Outcome and Impact

1. A verbal game assistant (via natural language typed via keyboard) that can be consulted during a game for both tactics as well  as analysis. [Q3]
2. Counterfactual reasoning and analysis assistant. E.g. ‘if I collected a ruby instead of the gold coin in level 2, would my options in level 3 be better or worse?’. [Q6]
3. Game summary generator, grounded in the rules and dynamics of the game. This could be customised as badges for sharing on social media. E.g. ‘I just won a thrilling game where my strategy to collect rubies trumped others who collected gold coins. In particular, this opened up the ability to …’. [Q10]
4. Customised Tutorial generator, that uses game logs to provide puzzles and tutorials of specific game states that were identified bottlenecks. [Q14]
5. A consolidated report and efficient codebase for all the above features [Q16] 

Data and Methodology

The above aims will be achieved via the following stages. 

1. Selection: Together with a supervisor from Tencent, we will identify a game (or a few games) for which the assistant will be developed.
2. Generate game logs: For each of the selected games, several game logs will be generated. This involves either playing the games, or simulating gameplay to generate logs of the game state. For example, if the game is chess, then the logs will contain the state of the board after each move has been made.
3. Reasoning engine: This engine will develop prediction models for optimal actions where possible or reasonable actions otherwise, based on the known state of the game.
4. Verbalisation engine: A representation of the game knowledge (say via a graph or simplicial complex) will be made accessible to a Language Model for ease of interpretation and verbalisation.
5. Generative modelling engine: The reasoning engine will be used to build a mechanism for generating arbitrary states of games, for tutorials or training material. For example, standard chess puzzles such as “white to mate within 5 moves” given a specific state of the board. Depending on the game, this might contain short, medium or long-term objectives.  

Students Requirements

• A good Bachelor’s degree (First Class Honours or international equivalent) or Master’s degree in a relevant subject
• Relevant research experiences a plus
• More on the CDT MLSystems requirements for candidates