Nanyang Technological University, Singapore | 24 August – 3 September 2026
What happens when technology leaves the classroom and meets the complexity of the real world?
For the PGP in Artificial Intelligence & Data Science cohort, the answer unfolded across two weeks at Nanyang Technological University, Singapore. The Study Abroad Module brought together digital transformation, technology strategy, machine learning, industry exposure and managerial judgement—moving students beyond how technology works to examine how and when it creates value.
The journey began with a deceptively simple question: what makes a technology capability worth having—and worth defending?
When Technology Becomes an Advantage
The opening sessions on Strategies for Digital Transformation with Prof. Anandasivam Gopal challenged students to look beyond technology itself.
Through the VRIN framework—Valuable, Rare, Inimitable and Non-substitutable—the cohort explored why some capabilities become lasting advantages while others remain easy to replicate.
The CEMEX case brought the distinction to life. Technology had transformed the cement company's logistics operations, enabling real-time delivery decisions and tighter customer commitments. Yet when CEMEX attempted to commercialise that IT capability outside its own domain, it faced established technology players and could not reproduce the same advantage.
The lesson was clear: technology creates lasting value when it is embedded in domain knowledge, processes, culture and context.
From Technology to the System Around It
The conversation then expanded from individual capabilities to the ecosystems that make them work.
Students explored technology stacks, cloud computing, ERP, CRM, Big Data, platform economics and network effects. Four decades of gaming-console history brought these ideas to life—from Atari and Nintendo to Sony, the Wii and OnLive—showing how market success depends not only on technical capability, but also on users, developers, licensing, business models and networks.
Technology investment added another dimension. Through the Whirlpool Europe case, students examined how uncertainty and flexibility can change the value of an investment. A project that appeared unattractive under conventional assumptions could become valuable when treated as a pilot with an option to scale.
The takeaway: technology decisions are also business decisions.
When Ideas Meet Enterprise Reality
The classroom moved into the enterprise through industry visits, beginning with Atos Singapore.
Discussions around AI, digital sovereignty and agentic AI showed how emerging technologies are being translated into organisational applications. One example demonstrated the potential of applied AI: a manufacturing quotation agent reduced turnaround time from 24 hours to approximately two minutes at 99% accuracy.
The cohort also heard from Andreas M Walter, Mahesh Enukollu and Jan de Meijer, connecting technology deployment to wider questions of security, governance and business value. At the Hyundai Motor Group Innovation Center Singapore (HMGICS), students encountered smart manufacturing through automation, sensors, VR and automotive testing. At Standard Chartered Bank, AI was examined through the lens of banking and organisational transformation.
Different industries. Different technologies. One recurring question: how does an organisation turn capability into measurable value?
The Right Model Begins with the Right Problem
The module then moved deeper into AI and Business Applications with Asst Prof. Vivek Choudhary and Prof. Vivek Chaudhary of NTU Singapore.
Students worked with regression, classification and clustering while exploring the bias-variance trade-off, correlation versus causation, A/B testing, MSE, RMSE and the complete machine-learning pipeline.
But the technical learning was consistently anchored in one principle: The problem comes first.
The Brew Haven Coffee case demonstrated how model selection, statistical interpretation and business judgement need to work together. Logistic regression introduced another important lesson: accuracy alone can be misleading. Target leakage, class imbalance, confusion matrices, precision, recall, F1 and balanced accuracy showed why a model must be evaluated in the context of the decision it is intended to support.
CRAW—Critical thinking, Reading, Analysing and Writing—and CCC—Concise, Clear and Coherent—reinforced the human side of analytical work. A model can produce an output. Turning that output into something useful requires interpretation, questioning and communication.
Two Weeks. One Larger Understanding.
By the final academic day, the cohort had moved through decision trees, CART and clustering, but the central question had remained remarkably consistent: what makes an AI solution useful?
A case involving an Indonesian wood-substitute product offered a particularly clear answer. The product was technically strong and supported by good data, yet its commercial success was constrained by a weak installer network. Data could identify symptoms; understanding the real cause required engaging with customers and the ecosystem around the product.
A guest session with Dr Dhruv Gupta, Inspector General of Police, Chhattisgarh, offered another perspective. His example of OCR being used to digitise handwritten FIRs illustrated that AI does not have to be sophisticated to create value. It needs to be matched to a real problem.
That idea became the thread connecting the nine-day journey.
The Study Abroad Module was never simply about learning more technologies or more models. It was about learning to question the problem, understand the context, evaluate limitations, communicate the output and connect technology with a meaningful outcome. At NTU Singapore, the cohort encountered technology through classrooms, case studies, enterprises and conversations with practitioners. They returned with more than a technical toolkit: a broader understanding of what it means to create value with AI and data science.
Because the future of AI will not be shaped only by those who can build models. It will also depend on those who know which problems are worth solving—and what to do when the model has given them an answer.