Introduction to AI 2400-MIKRO-AIBF-IIA

The course introduces participants to the fundamental concepts, practical applications and broader implications of artificial intelligence. It is organised into three interconnected blocks combining theoretical explanations, research-based examples, discussion and hands-on exercises.
The first block begins with an introduction to artificial intelligence, language models and foundation models. Participants learn how contemporary language models process information, generate responses and adapt their outputs to the context provided by the user. Prompting is presented as a method of placing a model in a temporary, task-specific state that can be shaped and controlled through instructions, context and examples.
Participants are introduced to a range of prompting techniques, including zero-shot and few-shot prompting, meta-prompting, chain-of-thought prompting, self-ask prompting and iterative prompting. Each technique is discussed in relation to its purpose, strengths and limitations. Relevant research findings are presented to show when particular approaches tend to improve performance and when they may be less reliable. Practical exercises allow participants to experiment with different formulations, compare model responses and refine prompts through iteration.
The second block focuses on the use of AI in programming and analytical work. Participants are introduced to OpenAI Codex as an agentic coding environment that can support the preparation, organisation and revision of analytical scripts. Using a common dataset and a set of introductory analytical questions, they configure the environment on their own computers and use prompting techniques to generate R code.
The code produced with AI is subsequently run and verified in RStudio. This workflow highlights the distinction between generating code and producing a correct analytical result. Participants are expected to inspect the generated scripts, identify possible errors, assess assumptions and verify whether the output answers the original analytical question.
This block also introduces the grammar of graphics and the basic principles of data visualisation in ggplot2. Participants learn how data, aesthetic mappings and graphical layers are combined to create visualisations. Particular attention is paid to selecting an appropriate graphical form, presenting information clearly and ensuring that a visualisation communicates a meaningful and accurate story about the data. The practical workshop allows participants to use Codex and RStudio to develop, run and improve their own analytical and visualisation workflows.
The third block examines the ethical, legal and social dimensions of artificial intelligence. Topics include the environmental costs of training and operating AI systems, the distinction between human augmentation and human replacement, manipulation and dishonest uses of AI, malicious and adversarial attacks, and responsibility for decisions or outputs produced with the support of AI systems.
The course also addresses privacy and personal data protection, the use of copyrighted materials in model training, and questions concerning the ownership and legal status of AI-generated content. Participants are encouraged to consider not only what AI systems can do, but also who is responsible for their use and consequences.
The final part of the course focuses on human-in-the-loop approaches and critical thinking. Participants discuss where human verification remains essential, where it may become a largely formal requirement, and how responsibility should be divided between people and automated systems. They consider how AI can be used to support reasoning without replacing it and reflect on how to maintain analytical, creative and critical-thinking skills in an environment where AI increasingly performs tasks previously completed by humans.
A central question throughout this part of the course is how people use the time saved by AI. Participants are encouraged to treat increased efficiency as an opportunity for deeper interpretation, verification and reflection rather than simply as a way to produce more output.

The project ‘Application of Artificial Intelligence in Business and Finance’ (No. BPI/SPI/2024/1/00078) is implemented by the University of Warsaw as part of the Spinaker programme organised by the National Agency for Academic Exchange (within the project ‘Wsparcie tworzenia i realizacji międzynarodowych programów kształcenia’, funded by the European Funds for Social Development 2021–2027).