By Anbu Muppidathi, CEO of Qualitest
There was a time when you had to know a programming language and a bunch of "command-line" code to operate a computer. Then came the visuals through graphical interfaces that revolutionized the computers with clicks and later "touch-screen technology to operate the devices. In the past few years, voice controls and gesture controls, and soon "thoughts" in the brain-computer interfaces are making the technology more human-like. The evolution makes the technology more intuitive, democratized, and accessible to all.
Every time an impactful technology is introduced, businesses reinvent themselves and in turn create a new normal for life and work. We call it a paradigm shift. These shifts challenge the status quo, introduce newer ways for production and consumption of resources, declare new winners and losers, and finally become the new normal … until a new theory emerges to challenge the existing paradigm. One such shift that is happening right in front of our eyes is AI.
Remember when the internet and cloud were introduced, it created new models of business, technology, process, and people. It fundamentally disrupted our lives and work. Businesses that resisted it met their demise and those that adapted survived.
As the world continues to debate on the productivity promises of AI with regard to development, testing, and operations, I believe we must shift our attention to a larger opportunity before us: AI becoming the major interface for all applications.
Let’s compare the adoptions of cloud and AI technologies. Though cloud benefits are well-known, only 20% of the workloads in the corporate world has moved to cloud so far, mainly infrastructure, client-facing applications, and data. This is due to challenges such as maintaining a hybrid environment, transitioning management from the 'old' to the 'new', inconsistencies arising from the change, and a lack of in-house management skills and tools, talent acquisition and administration, and so forth. Many customers “lifted-and-shifted” to quickly move to cloud and some have built cloud-native applications and environments.
AI adoption is expected to follow a gradual trajectory, where many will just lift and shift to take advantage of the ease of using AI to improve CX. Subsequently, a shift toward AI-native development will occur, allowing for the comprehensive utilization of AI's capabilities. Regardless of the path chosen, it is imperative for the development, testing, and operations (dev-test-ops) community to upskill to handle the AI-challenges.
- Changing design patterns: Design patterns are well known solutions for the most common problems that occur while designing something. They save time and make code easier to understand and faster to develop and debug. AI adoption and AI applications will demand new design patterns.
- Integration complexities: Emergence of APIs and UIs that connect AI and the existing tech assets (applications, data, devices, and so on) will introduce a new stream of innovation, design, development, and test complexities.
- Importance of Ground Truth to understand the user personas: AI allows more human attributes in the interaction than any other technology, which forces the fine-tuning of the AI applications in handling the human interactions much more than the applications of the past. For example, the culture of the demographics, the accessibility and localization of the consumers in the AI interaction, and so forth are critical components in training the LLMs.
- Testing complexities: GenAI is subject to continuous testing and human oversight. So-called “hallucinations, which refers “incorrect or misleading” results that AI/ML models generate, occur if there are flaws in training data or the model design, otherwise known as “design error,” or the training data that we used to train the AI/ML model may be wrong, known as “garbage-in-garbage-out.” Validating the training data against the real-world usage data and reviewing the model architecture for quality will unearth hallucinations, bias, and errors. Adding AI performance and security in the mix will make testing hyper-complex. We need AI to test AI.
- AI operations: Business operations will shift dramatically when AI models come into play. Continuous model quality, performance, and security will demand hyper-automation of validating for these, at speed. Processes for error-tolerance, error-handling, and model (re-)training, and so on should all be through before AI models take the driver seat.
Every time a new technological paradigm is introduced, the skills gap in the workforce
widens further. Reskilling and upskilling are hygienic factors taken for granted, and given the lack of upskilling amongst individuals it is now becoming critical, especially in the age of Generative AI. So, stop worrying about whether AI will kill jobs, and start to think about what it takes to upskill to face the new normal. Unless we elevate our thinking beyond software development lifecycle (SDLC) layers, we cannot elevate the dev, test, and ops discipline to the demands of AI. The ones that are experienced in the current systems are the ones that can easily step up to the new paradigm.
About Anbu
Anbu Muppidathi is the CEO of Qualitest. A technology veteran with more than 30 years of experience in digital transformation and technology modernization, Anbu has world-class operational and go-to-market expertise. Before joining Qualitest, Anbu most recently served as Global Head of Cognizant’s Enterprise Cloud Application Services. Prior to that, while running Cognizant’s Quality Engineering and Assurance practice between 2014 and 2018, he more than doubled the company’s testing revenue to $2.2B in annual sales with a team of 35,000 professionals while improving its analyst rankings to the leader status.
The views and opinions expressed herein are the views and opinions of the author and do not necessarily reflect those of Nasdaq, Inc.