Topic 7

Lesson 7. Application of Artificial Intelligence for Data Analysis. Copilot, ChatGPT, Google Gemini. Prompt Engineering for Analytics: Formula Generation, CSV Array Interpretation, Ethics, and AI Hallucination Checking

Objective: formation of students with a complex of theoretical knowledge, critical engineering thinking and practical skills in the field of applying generative artificial intelligence technologies for data analysis, development of deterministic analytical prompts, automated interpretation of tabular arrays and ensuring cyber hygiene and verification of results within the concept of AI-Augmented Analytics. The student must master the evolution of analysis paradigms from low-level programming (Pro-Code) to Prompt-Driven analytics, architectural components of augmented analytics (Augmented Data Prep, Discovery, code generation, NLG storytelling), as well as comprehend the probabilistic nature of large language models (LLM) and the mechanisms of the emergence of mathematical and logical hallucinations. During the training, special attention is paid to the comparative analysis of leading cloud ecosystems - distinguishing between the Python Sandbox (Code Interpreter) in ChatGPT, the multimodal environment with a large context window in Google Gemini, and enterprise integration with Power BI and DAX in Microsoft Copilot. Students should learn how to practically design structured queries using the R-C-T-O (Role – Context – Task – Output) framework, apply specialized Chain-of-Thought and Few-Shot Prompting techniques to generate complex SQL queries, QUERY functions, and DAX dimensions, as well as configure the work of autonomous AI agents for step-by-step profiling, cleaning, statistical analysis (EDA), and visualization of CSV arrays. An important analytical and evaluation outcome is the ability to implement a strict four-step anti-hallucination protocol (enforcement of code execution, validation on microsamples, cross-validation, Human-in-the-Loop principle) and ensure compliance with data privacy standards (GDPR, ISO/IEC 27001) through anonymization, PII masking, and the use of secure Enterprise API or Self-Hosted models for safe and informed management decision-making.