Ezanne Grobler, Head of Quality Assurance competency, iOCO.
iOCO has launched one of the first locally developed ISTQB-aligned AI testing e-learning courses in South Africa, to upskill quality assurance testers on the latest best practices in AI systems testing.
Designed by iOCO subject matter expert Aysha Sassman, the course is aligned with the International Software Testing Qualification Board (ISTQB) Certified Tester-AI Testing (CT-AI) syllabus, which is the globally recognised benchmark for testing professionals.
iOCO is in the process of having the Testing AI Fundamentals course accredited by SASTQB.
Meeting a need for AI-specific testing skills
The new course addresses a growing market need for AI-specific training for testers, developers and business stakeholders, says Ezanne Grobler, Head of Quality Assurance competency at iOCO.
Grobler notes: “Traditional testing skills alone are not enough for AI testing; you also need additional skills, characteristics and techniques. In many AI systems, you don’t have a single, deterministic expected result per input, and behaviour can change over time as models are retrained or data drifts. Hence, we need different ways to define oracles and acceptance criteria. The main difference between traditional systems and intelligent systems is the mere fact that in traditional systems you have an expected result, so you can validate and verify the accuracy of the system based on the expected result. In many intelligent systems, you don’t have a single deterministic expected result per input, and when models are retrained or the data distribution drifts, behaviour can change over time. So, testing an intelligent system is a totally different ball game.”
She elaborates: “In traditional testing, you will use techniques like boundary value analysis or equivalence partitioning because you have a clear specification and can predict the expected result. With AI systems, your testing techniques must also look at things like bias, hallucination and model or data drift. Because these systems learn from data and are often retrained or used in changing environments, their behaviour can slowly move away from the original intent if you don’t actively monitor and govern the data and feedback they learn from.” She adds that understanding quality assurance for AI systems is crucial for QA testers, developers, data analysts and business stakeholders alike.
Comprehensive course content
iOCO’s new Testing AI Fundamentals e-learning course is designed to make AI and AI quality assurance and testing practical and understandable.
Grobler explains that the course was developed over a year, initially to upskill internal staff. “However, we realised it was such a comprehensive and important guide that we elected to make it available as a self-paced e-learning course for the broader IT and business community,” she says.
“We have done a lot of training and lecturing, but this is our first e-learning course,” she says. Grobler believes the course is unique as an e-learning offering because it includes exclusive access to a private community forum, where learners can connect with like-minded professionals working in AI testing. She expects interest from across South Africa, the rest of Africa and even internationally.
At recent meet-ups in Johannesburg and in Cape Town, a soft launch of the course was well-received, with the majority of delegates registering their interest in participating in the training.
“While our internal testers who have completed the training concede that it’s a tough course, they responded positively to the use of case studies, visuals and videos to transform theory into practical, testable knowledge. The course equips participants with the skills they need to assess AI quality and reduce AI-related risk for the organisation,” she says.
Grobler adds: “Business can benefit from this course too, as it gives insight into the layers of artificial intelligence systems and helps them understand AI-related risks, the governance required around AI and how to mitigate those risks through testing, validation and verification.”
The course requires no prior AI or QA experience and covers the basic concepts of AI, how quality is defined in AI and how to apply modern techniques such as adversarial, back-to-back, metamorphic and exploratory testing.
While the course is self-learning and self-paced, Grobler estimates it involves around 40 hours of intense learning. The learning journey includes modules such as an introduction to AI, quality characteristics of AI systems, machine learning fundamentals, testing AI-based systems, data handling and performance metrics. Learners will gain practical insights through interactive content and quizzes, and will gain a holistic understanding of AI from a business and technical perspective, within a globally recognised framework for testing and assuring AI systems. Completing the course also puts participants on a pathway to the ISTQB CT-AI certification.
Grobler says: “Even though we don’t have ISTQB certification yet, we align to the syllabus, so as a result, once people complete our course, they have a strong foundation for writing the ISTQB exam, barring that you already hold the ISTQB Foundation level (CTFL) certificate.”
The course will be updated continuously as AI evolves and as the ISTQB syllabus changes. Grobler says: “Because the AI environment is continually evolving, ISTQB has recently launched a new specialist syllabus (CT-GenAI) on using generative AI to augment software testing. In the same spirit, one can expect to review and refresh the AI-related course content more frequently than traditional testing content, which typically changes less often.”
iOCO is currently running a launch special on the Testing AI Fundamentals e-learning course. The first 100 people to register will receive a 50% discount, allowing them to enrol for only R2 475.
Click here to register. Click on the “login” button, scroll down to “create new account” and follow the process to the point of payment.
