Skip to main content

Testing AI Chatbots for Real

6-hour Masterclass

AI chatbots are not impossible to test well as black-box systems. What does not work is treating testing as a few prompts, a quick glance at the answers, and a leap of faith.

Timetable

8:30 a.m. – 4:30 p.m. Monday 16th

Room

F-,E- & D-Rooms

Artificial Intelligence (AI)

Audience

Testers, Developers, Product people. No AI or machine learning background is needed.

Required

Laptops with Windows Remote Desktop Access Tools pre-installed

Key-Learnings

  • Learn practical ways to test chatbot behavior from a black-box perspective.
  • Find useful ways to assess response quality even when answers are subjective, partial, or variable.
  • Use structured exploration to expose risks, unfair behavior, and unexpected failure patterns.

Stop Guessing. Test the Bot.

AI chatbots challenge many of the assumptions behind traditional software testing. From a user point of view, the real question is simple: how do we build confidence in what people actually experience? Chatbots are conversational, context-sensitive, and often variable in their responses, which makes them hard to judge with a simple pass/fail mindset.

In this full-day tutorial, we take a practical black-box approach to testing AI chatbots. We start by building a basic understanding of how chatbot behavior is shaped by prompts, memory, tools, rules, and surrounding application logic. We then focus on what can be observed and tested from the outside: responses, conversation flows, boundaries, integrations, fallback behavior, and visible failure modes.

From there, we apply classic test design techniques and look at practical ways to evaluate quality. In the final part, we use exploratory and metamorphic testing to uncover language-specific risks such as ambiguity, manipulation, and hidden weaknesses.

Related Sessions

There are currently no related sessions listed. Please check back once the program is officially released.