- BTW media recently had the opportunity to speak with Gokul Choudhary, an experienced QA engineer currently working at Tech Mahindra.
- Recently, he has been focusing on AI testing, working on innovative projects at Meta (formerly Facebook) that involve AI-driven functionalities for smartwatches and smart glasses.
- His deep experience with leading social media platforms like Facebook, WhatsApp, and Instagram, combined with his Agile software testing certifications, has made him a key figure in the evolving landscape of software and hardware QA.
- Tech Mahindra, a prominent player in the global IT services and consulting industry, is renowned for its commitment to innovation and excellence.
- The interview with Gokul Choudhary offers a compelling glimpse into the evolving role of Quality Assurance (QA) in the realm of artificial intelligence.
As AI becomes increasingly integrated into everyday technology, the role of Quality Assurance (QA) engineers is also shifting.
We recently had the opportunity to speak with Gokul Choudhary, an experienced QA engineer currently working at Tech Mahindra , who has spent over a decade in software testing and more than two years specialising in AI-driven applications. With over a decade in software testing, Gokul Choudhary has a diverse QA background, covering mobile, desktop, and web applications. His expertise spans e-commerce, healthcare, banking, and voice applications. Recently, he’s concentrated on AI testing at Meta, working with smartwatches and smart glasses to ensure accurate AI commands and responses. His extensive experience with major social media platforms and Agile certifications positions him as a significant player in the evolving QA landscape.
Q: Today, we would like to discuss a few topics, including the differences between traditional software testing and AI application testing, as well as the tools and techniques you find most useful. To start, could you provide an overview of the key distinctions between testing traditional software and AI-driven applications? How does the presence of AI influence your testing strategies?
Gokul:
Sure. When it comes to AI, let’s take the example of smart glasses. When we give a command, like asking, “What is the weather now?”, the AI connects to a database or an API to fetch the response based on the current location. We test the basic functionalities to ensure that the AI responds within a given time frame.
With traditional software, you might need to type commands or inputs manually. However, AI allows for voice commands, which makes it much faster and more convenient, especially when you’re on the go or unable to type. This is one of the key advantages of AI—it speeds up processes and enhances user experience by providing quick responses in real-time, but it lacks the intuition to understand context and nuance.
Q:So, AI makes the process much faster and more efficient. What tools and techniques do you find most effective for testing AI-driven applications? Could you share some examples?
Gokul:
Actually, for AI testing, we don’t rely on any specific automated tools. Most of the testing is done manually. We work with Generative AI, like prompt-based AI models, and use natural language commands to interact with the AI.
For example, in Meta’s AI, we activate the device and issue commands. The AI processes these commands and responds accordingly. It also stores voice commands and their responses in a chat format within the connected mobile app. This allows us to review and cross-check all interactions.
Q: So, you’re using Meta AI primarily, which sounds similar to ChatGPT or Gemini. Is that right?
Gokul:
Yes, that’s right. While I do use ChatGPT occasionally for testing purposes, Meta AI has more specific capabilities. For example, when we issue a command to Meta AI, it stores all voice commands and responses in a chat history on the mobile app. This enables us to track and review all interactions.
For instance, if I give a command to capture an image of a building using smart glasses, Meta AI processes the image, identifies the building as a shopping mall, and provides information about it. All this data is stored in the chat server and is accessible in the chat history.
Q: That’s fascinating. Given the rapid advancements in AI, it’s clear that there are many benefits, but also some challenges, such as ethical considerations. Could you elaborate on that?
Gokul:
Absolutely. The ethical considerations are indeed unavoidable, especially with the extensive data collection and AI’s growing autonomy.
Q: Could you clarify what you mean by ethical considerations?
Gokul:
Sure, there’s a tendency to view AI as a panacea, but it’s merely a powerful tool. Many organisations rush to adopt full AI automation only to discover gaps where AI’s capabilities fall short. For instance, AI-driven testing tools may struggle to accurately interpret creative design elements or fail to account for culturally specific user behavior, leading to a misalignment with user needs. This is where the human touch becomes invaluable.
When working with AI, it’s crucial to ensure data privacy and user consent are respected. We must be transparent about how data is used, stored, and shared to prevent any misuse of AI technologies. This involves careful planning and adherence to ethical guidelines to protect user rights and data integrity.
AI’s role in QA will continue to grow, but it will always require human guidance, especially as new and unpredictable challenges emerge. Although AI is an incredible ally, but the human mind is irreplaceable.
About Tech Mahindra
Tech Mahindra, a prominent player in the global IT services and consulting industry, is renowned for its commitment to innovation and excellence. Established in 1986 as part of the Mahindra Group, Tech Mahindra has evolved into a leading provider of digital transformation solutions, IT services, and business process outsourcing.
Personal opinion: Navigating the evolution of QA in the age of AI
The interview with Gokul Choudhary offers a compelling glimpse into the evolving role of Quality Assurance (QA) in the realm of artificial intelligence. Gokul’s extensive experience underscores a critical shift in the industry—while traditional software testing relied heavily on manual inputs and pre-defined procedures, AI introduces a new dimension of testing that is both dynamic and complex.
Shift from manual to voice-activated testing: Gokul’s insights into AI testing reveal a fundamental transformation in how we approach QA. Traditional testing often involved typing commands and validating results manually. However, AI-driven applications, particularly those using voice commands like Meta’s smart glasses, streamline this process. The ability to issue voice commands and receive real-time responses not only enhances efficiency but also aligns with the fast-paced demands of modern users. This shift highlights a broader trend in technology—where convenience and speed become paramount.
Manual testing in an automated world: Interestingly, despite the advancements in AI, Gokul points out that testing AI applications still largely relies on manual techniques. This seems counterintuitive in an era where automation is ubiquitous. The reliance on manual testing for AI-driven features, such as voice interactions and real-time data processing, emphasises the need for human oversight in validating AI performance. It’s a reminder that while technology evolves, the human touch remains crucial in ensuring quality and accuracy.
Ethical considerations: The discussion about ethical considerations is particularly noteworthy. As AI technologies handle increasingly sensitive data, ensuring data privacy and user consent becomes paramount. Gokul’s emphasis on transparency and adherence to ethical guidelines reflects a growing awareness of the responsibilities that come with AI development. The challenge of balancing innovation with ethical standards is a significant concern that will shape the future of AI and its integration into daily life.
