What are artificial intelligence and ChatGPT?

Artificial intelligence in general

Artificial intelligence (AI) is a field of computer science that seeks to create systems capable of performing tasks that previously required human intelligence—such as understanding language, making decisions, learning, or solving problems. AI solutions use mathematics, statistics, machine learning, and deep learning to model and process complex patterns in data.

AI systems can be of different types: narrow artificial intelligence (narrow AI), which is designed for a specific task, and artificial general intelligence (strong AI / AGI), which could theoretically solve a broad range of tasks at a human level—currently, general AI does not actually exist.

What is ChatGPT?

ChatGPT is one example of a generative language model (large language model, LLM). It is based on a large language model trained on a vast amount of text to understand and generate human-like language. ChatGPT can answer questions, write essays, translate, explain complex concepts more simply, assist with coding, etc.

How ChatGPT works in brief:

1. Training Phase
   Models are trained on large text corpora—from books, articles, web content, etc. The model learns to predict which word is likely to follow a sequence of preceding words.

2. Fine-tuning
   After general training, the model may be tailored to perform more specific tasks, for example by training it to answer questions or follow specific style rules.

3. Processing input and producing output
   The user provides a prompt (a question, instruction, or the beginning of a text), the model calculates the probabilities of the next words and generates a response. Complex calculations, attention mechanisms, etc. operate in the background.

Although ChatGPT often appears to speak in natural language, the model does not actually “know” things like a human does—it does not understand everything about the world but instead derives patterns from training data. Therefore, its answers may contain errors or inaccuracies.

Capabilities, advantages, and limitations

Capabilities and advantages

• Speed and versatility
  ChatGPT can generate text very quickly, answer questions, help generate ideas, and structure texts.

• Language processing and multilingual assistance
  It can translate texts, explain complex concepts in simpler language, and help with grammar and style errors.

• A supporting role in learning and work processes
  Suitable for visualizing ideas, planning tasks, preparing research papers, and initiating analyses.

• Flexibility and accessibility
  ChatGPT can be used in various environments (web applications, integrated APIs) and does not require constant human supervision.

Limitations and risks

• Inaccuracies and “hallucinations”
  The model may generate information that appears logical but is factually incorrect.

• Lack of transparency
  It is difficult to consistently trace how the model arrived at a specific answer, which reduces reliability in sensitive contexts.

• Dependence and decline in critical thinking
  Excessive reliance on ChatGPT may reduce learners' own analytical and independent thinking skills.

• Authorship and plagiarism
  If a pattern or large texts are generated without clear disclosure and presented without attribution, ethical or legal issues may arise.

• Ethical and regulatory aspects
  Educational institutions and organisations may establish rules for using ChatGPT to prevent dishonest practices. hm.ee+2hm.ee+2

For example, Tartu Ülikool has guidelines for using chatbots in teaching. Tartu Ülikool Transparency has also been recommended in educational institutions: users should indicate when ChatGPT has been used. Tartu Ülikool+1

Uses of ChatGPT in studies and work

Below, I describe the areas where ChatGPT can be the most useful tool:

Teaching and academia

• Idea generation and literature review
  To begin, you can ask ChatGPT for a list of topics or guidance on researching the literature.

• Creating a structure
  For example, it helps generate a draft article or essay — an introduction, sections, and conclusions.

• Explanations of complex concepts
  Mathematics, physics, economics, etc. – ChatGPT can provide explanations at different levels of difficulty.

• Language Assistance and Editing
  Help with correcting grammatical errors, adjusting style, and making text clearer.

• Language translation and multilingual learning
  Translating between different languages or explaining concepts in multiple languages.

• Homework and problem-solving
  For example, instructions for solving math problems, code examples, or brainstorming.

• Generating tests and questions
  Teachers can request the generation of open-ended or multiple-choice questions on various topics.

Such use is already relevant in education: for example, schools use ChatGPT to improve learning materials or provide feedback. Areng.ee+2raatuse.tartu.ee+2

Work and Professional Tasks

• Content marketing and blog writing
  Helps generate ideas for initial drafts, titles, paragraphs, and SEO phrases.

• Correspondence and documents
  Drafting emails, reports, summaries, and project plans.

• Data Analysis and Visualization Ideas
  Although ChatGPT does not visualize data, it can help find suitable approaches and explain the methodology.

• Customer support and interactive communication
  Creating chatbots and automated responses for frequently asked questions.

• Writing code and providing guidance
  Help with programming languages, examples, debugging, or explanations.

• Language correction and support
  Proofreading, stylistic changes, word choice, translations.

• Improving ideas and revisions
  If a draft text already exists, ChatGPT can rephrase or expand it.

Professionally, it is particularly suitable for those with many text-related, analytical, or writing responsibilities.

A practical approach—how to get started

Although the main part of the article does not provide direct recommendations, I will present a general model here for how its use could be approached:

1. Define your goal
   Think about what you want to use ChatGPT for (learning, generating ideas, drafting documents, etc.).

2. Create a clear prompt
   Describe exactly what you expect: format, length, style.

3. Analyse the response critically
   Do not automatically accept the output at face value — verify the facts, logic, and reliability.

4. Iterative approach
   Adjust and refine the prompt as needed to get better matches.

5. Document sources / cite
   If you use text excerpts from ChatGPT, mark or cite them (as in academic work).

6. Fine-tune the workflow
   Integrate ChatGPT into your regular workflow — either as a separate step or as an additional tool.