Show all
Log in
0
Basket
0,00 €
Log in
0
Basket
0,00 €
Pumpedu Artificial intelligence (AI) Generative AI in a nutshell: how models really work

Generative AI in a nutshell: how models really work

The module explains the principles of generative AI without unnecessary technical burden – why models respond in a certain way, why they can make mistakes, and why a smooth response is often mistaken for the truth. Participants will understand when to validate output and why.
Level
Designed for participants without knowledge and experience
basic
Course length
2 hours
Language
cz
Course code
PU00010039
Select a date 
Artificial intelligence (AI)
Category:
Artificial intelligence (AI)
Do you want this tailor-made course to your company? Contact us

Courses with lecturer

Term
Language
Place
Form
?
How and where the course takes place.
Price without VAT
21. 8. 2026 09:00 - 12:30
Language
Place
online
Form
virtual classroom
?
Online training with a lecturer at a specific time.
Code of the course: PU00010039-0002
Price without VAT
3 900 Kč
Order
15. 10. 2026 09:00 - 12:30
Language
Place
online
Form
virtual classroom
?
Online training with a lecturer at a specific time.
Code of the course: PU00010039-0003
Price without VAT
3 900 Kč
Order
4. 12. 2026 09:00 - 12:30
Language
Place
online
Form
virtual classroom
?
Online training with a lecturer at a specific time.
Code of the course: PU00010039-0004
Price without VAT
3 900 Kč
Order
Open term
?
We will agree on a specific date together. This is a non-binding order.
Language
Place
online
Form
virtual classroom
?
Online training with a lecturer at a specific time.
Code of the course: PU00010039-0001
Price without VAT
3 900 Kč
Order

Course description

Generative AI tools are convincing – they respond fluently, confidently and often very quickly. But that doesn’t mean their answers are always correct. This module explains how language and generative models actually work, in a way that is understandable even without a technical background.
Participants will learn how a model works with probability, context and patterns in data, and why its answer differs from real knowledge or verified fact. The module explains why so-called hallucinations arise – i.e. convincingly formulated but inaccurate or made-up answers – and what role training data, the way the model is trained and the context the model is given play in this.
It also includes an orientation on how public AI tools differ from corporate and specialized solutions, and why this choice affects how much the output can be trusted. Participants leave the module with a clearer idea of ​​why output control is a necessary part of working with AI, not an unnecessary formality.

Target audience

  • employees and managers who want to understand the principles of how AI tools work, not just how to control them,
  • companies that want their team to be able to critically assess AI outputs,
  • people without a technical education who need an explanation without simplifying it into empty phrases,
  • participants following the first module of the AI ​​Academy.

Course content

  • What are language models and generative tools
  • How a model works with probability, context, and patterns
  • The difference between knowledge, estimation, and generated response
  • Why hallucinations and inaccuracies arise
  • The role of data, training, and context
  • The difference between public, corporate, and specialized tools

Certification

Upon completion of the course, you will receive a Pumpedu certificate issued by an authorized training provider with accreditations from leading international organizations in IT, project management, and professional development.
 
Obrázek certifikátu

Objectives

  • Understand how generative models work with probability and context.
  • Distinguish between knowledge, estimation, and generated response.
  • Understand why hallucinations and inaccurate outputs arise.
  • Understand the role of data, training, and context in generating a response.
  • Understand the differences between public, corporate, and specialized tools.

Frequently Asked Questions

What is a generative model?
A tool that creates new text, image, or other output based on probability and patterns in data – rather than searching for a ready-made answer.

Why does a model sometimes give an inaccurate answer, even if it looks convincing?
The model generates the most likely continuation of the text, not a verified fact. This phenomenon is called a hallucination.

What is the difference between knowledge and a generated answer?
Knowledge is verified information, a generated answer is a probabilistic estimate based on learned patterns – these two things may or may not coincide.

Why is the difference between public and corporate AI tools important?
They differ in what data they process, how they handle corporate content, and what level of control and security they offer.

Is the module suitable even without a technical education?
Yes, the principles are explained clearly and without the need to understand the technical details of how the models work.

Lecturers

Michal Kolomazník
Michal Kolomazník

Michal currently works at Microsoft as an Agile Coach & Principal Project Manager & PCAI, where he leads an AI transformation portfolio with a team of over 70 people. He was at the birth of the Business Value Program and for his work he received the Microsoft SPARK Award and inclusion in the Platinum Club (reserved for the 200 most influential people within Microsoft).

Michal is not “just” a technical specialist or “just” a manager, but he can connect both levels:

  • Technical experience with AI – implementing RAG and agent AI solutions from prototype to production operation, working with Copilot Studio, Azure AI and other tools
  • Change management – ​​certified Prosci Advanced Instructor, i.e. a person who trains and coaches other lecturers and leaders in change management – ​​exactly what is the most common weak point in AI adoption
  • Governance and security – ISO/IEC 42001 certified auditor (AI management systems), CISO certification, experience with Microsoft Sentinel and Defender
  • Business experience – previously Finance Manager at Procter & Gamble (CAPEX and investment planning management for a business unit with a turnover of over 1 billion USD) and Purchasing & Quality Manager at ENGIE
Thanks to this combination, Michal can truly adapt the training to the target group – sometimes he will lead a workshop for company management, sometimes technical training for the IT team and sometimes a practical course for regular users. It is not one universal slide deck, but content tailored to the role, experience and specific situations of the participants.

Do you want this tailor-made course for your company?

Contact us

News with the course

Náhledový obrázek novinky
Artificial intelligence (AI) 15. 11. 2024
The practical application of AI: HubSpot increased lead conversion rate

HubSpot is an American software development company. It has 8,000 employees and branches in several countries around the world. Processes related to lead generation and qualification were time-consuming and not always efficient, reducing sales productivity.

Náhledový obrázek novinky
Artificial intelligence (AI) 17. 10. 2024

AI in practice: DHL has accelerated the delivery of parcels

DHL is one of the largest logistics companies in the world. It provides transportation and delivery of parcels in many countries. However, route planning, warehouse management and demand forecasting were difficult and often inefficient.

Náhledový obrázek novinky
Artificial intelligence (AI) 12. 9. 2024
The practical application of AI: Microsoft increases financial prediction accuracy by 20%

Microsoft, a global tech giant, faced challenges with its complex financial processes, which involved extensive financial planning and analysis (FP&A). Traditional methods were time-consuming and sometimes inaccurate, leading to delayed decisions and potential financial risks.

More news

Previous courses

Do you want this tailor-made course for your company?

Contact us

News with the course

Náhledový obrázek novinky
Artificial intelligence (AI) 15. 11. 2024
The practical application of AI: HubSpot increased lead conversion rate

HubSpot is an American software development company. It has 8,000 employees and branches in several countries around the world. Processes related to lead generation and qualification were time-consuming and not always efficient, reducing sales productivity.

Náhledový obrázek novinky
Artificial intelligence (AI) 17. 10. 2024

AI in practice: DHL has accelerated the delivery of parcels

DHL is one of the largest logistics companies in the world. It provides transportation and delivery of parcels in many countries. However, route planning, warehouse management and demand forecasting were difficult and often inefficient.

Náhledový obrázek novinky
Artificial intelligence (AI) 12. 9. 2024
The practical application of AI: Microsoft increases financial prediction accuracy by 20%

Microsoft, a global tech giant, faced challenges with its complex financial processes, which involved extensive financial planning and analysis (FP&A). Traditional methods were time-consuming and sometimes inaccurate, leading to delayed decisions and potential financial risks.

More news

Why with us

Online training, classroom courses, e-Learning, webinars, etc.
We create custom courses that you won't buy anywhere else
Experienced team of experts
Course and staff management tool