1. Analyze a wide range of artificial intelligence methods and applications
2. Explain the basic principles of large language models
3. Explain the basic principles of generative models for images and 3D shapes
4. Explain the basic principles of reinforcement learning
5. Develop an application prototype using various artificial intelligence methods
6. Compare different artificial intelligence models in the context of limitations and challenges they present

Theoretical lectures

Project to create an application that applies artificial intelligence

Preparations for lectures

Downloading content related to course performance

1. Introduction and acquaintance with the course plan/objectives. History and overview of artificial intelligence applications in games, virtual environments, and engineering in general.
2. Classic artificial intelligence methods.
3. Fundamentals of machine learning and pattern recognition.
4. Neural networks and the backpropagation algorithm.
5. Introduction to large language models (LLMs) and chatbots. Prominent early examples (GPT-1 and 2, BERT) and modern large models (LLaMA, GPT-4, etc.). Evaluating accuracy and scaling laws. Instructional tuning.
6. Effective use of large language models in practice: decoding strategies, prompt engineering. Summarizing, expanding, transforming, and analyzing text. Acceleration and adaptation for practical applications: quantization, pruning, and distillation.
7. Advanced methods of using large language models: improvements with the application of text search (RAG) and the use of tools (toolformers), tuning, and the PEFT method, agent systems. Introduction to multimodal models.
8. Midterm exam
9. Generative models for images and 3D shapes.
10. New representations of virtual scenes: neural radiance fields and Gaussian splatting.
11. Introduction to AI agents. Agents based on large language models.
12. Fundamentals of reinforcement learning.
13. Fundamentals of reinforcement learning -- continued.
14. Limitations and challenges of modern artificial intelligence. AI safety. Conclusion.
15. Final exam.