Organizations across Asia have demonstrated a remarkable proficiency in deploying artificial intelligence (AI) technologies. However, the critical question remains: are these sophisticated tools being utilized effectively to drive tangible business outcomes? Jun Tay, a finance technology specialist at DBS Bank in Singapore and a member of the ACCA (Association of Chartered Certified Accountants), posits that the crucial differentiator between mere deployment and successful AI adoption lies in bridging the gap between technological capability and practical application. This chasm, she argues, is where the true success or failure of AI initiatives is determined.
"AI adoption most often breaks down at the onset of change: the point where operating reality, data context, and human behavior meet," Tay explained in a recent discussion. While acknowledging that capable AI platforms are readily available, Tay emphasized that adoption only becomes truly meaningful when end-users understand the specific problem the AI is designed to solve, develop trust in the underlying knowledge base powering the tool, and possess the skills to effectively integrate its outputs into their daily workflows.
This perspective challenges the conventional notion that AI adoption is primarily a technological hurdle. Tay believes AI’s distinctiveness from previous waves of workplace technology lies in its accelerated exposure of organizational inefficiencies and skill gaps. AI doesn’t just alter the systems people use; it fundamentally reshapes how they formulate inquiries, interpret information, and ultimately make decisions. The required shift for employees is substantial: moving from manual data retrieval to sophisticated prompt engineering, from relying on static documents to critically validating AI-generated responses, and from viewing technology as a separate entity to seamlessly embedding it into the operational fabric of their work. "That transition requires confidence, ownership, and repeated practice," Tay stated, underscoring the human element in AI integration.
Insights from a Generative AI Pilot in Finance
Tay’s observations are deeply informed by her experience supporting a generative AI chatbot pilot within a finance function. The underlying technology was already accessible and scalable, requiring only the integration of a relevant knowledge base. However, the pilot revealed that the most significant challenge was not technical implementation but the effort required to make that knowledge truly usable for the finance team.
During her review of the pilot’s queries, Tay identified a key determinant of answer quality: the precise alignment between a user’s prompt and the language used in the underlying documentation. To enhance accessibility for a broader user base, she initiated efforts to convert knowledge files into more reliably readable formats. While the pilot ultimately achieved an impressive 89.8% success rate, exceeding the recommended 80% threshold, Tay’s primary takeaway was not the raw performance metric. "Adoption depended on knowledge, design, access, governance, and user readiness, not just the AI engine," she asserted.
The pilot also highlighted what Tay describes as a "translation gap," an early indicator that a workforce may not be fully prepared for the new tools being introduced. For instance, a user might query about "milestones," while the relevant document refers to an "operating cycle." This linguistic discrepancy can lead to suboptimal AI responses, not due to a technological failure, but because of a misalignment between business terminology, documentation structure, and the AI’s retrieval mechanisms. "That is not a technology failure; it is a translation gap between business language, documentation structure, and AI retrieval behavior," Tay clarified.
Identifying and Addressing AI Adoption Warning Signs
Beyond the translation gap, Tay identified several other recurring warning signs that signal potential AI adoption challenges. These include employees reverting to familiar, pre-AI workflows due to uncertainty about when to trust AI outputs, the submission of low-quality prompts, inconsistent interpretation of AI-generated information, an over-reliance on AI without critical validation, or outright avoidance of the tool because users fail to perceive its relevance to their specific responsibilities. In the finance sector, where outputs frequently inform critical reporting, control reviews, and strategic stakeholder decisions, the stakes associated with these issues are significantly elevated.
Tay attributes these pervasive challenges to an underinvestment in change management. She observed that organizations tend to concentrate their resources and efforts on the technological rollout itself, allocating insufficient attention to crucial aspects such as detailed use-case design, comprehensive prompt guidance, the quality and structure of documentation, robust training programs, and effective governance frameworks. Her recommended approach is to treat AI adoption as a fundamental operating model transformation. This involves clearly defining the scope of the tool’s application – what it should and should not be used for – establishing clear ownership of the knowledge base, implementing rigorous quality review processes, and fostering user confidence through consistent and positive experiences.
However, Tay is careful not to label initial imperfections as outright failures. "Most products and tools are not released in a perfect or ideal state," she commented. "From a progress standpoint, starting, failing safely, and learning quickly are far better than avoidance or remaining with the status quo." This pragmatic outlook emphasizes the iterative nature of AI integration and the importance of continuous learning and adaptation.
The Enduring Role of Human Accountability in AI
A cornerstone of Tay’s philosophy on responsible AI adoption is the unwavering principle of human accountability. "When AI fails, creates risks, or leads to a poor decision, humans remain accountable. You cannot sue an AI in the courtroom," she stated unequivocally. "AI may support judgment, but it does not replace responsibility."
This fundamental principle underscores the indispensable role of domain professionals in the AI adoption process. While technologists are responsible for building the AI platforms and enabling the underlying capabilities, it is the domain experts who possess the critical acumen to assess the meaningfulness of AI outputs, the soundness of their interpretation, and the appropriateness of any recommendations within their specific operational context. In the finance pilot she supported, Tay’s contribution was not about technically tuning the AI model itself. Instead, her focus was on evaluating whether the AI’s responses were logically coherent and practically useful to enterprise users grappling with real-world financial questions. This involved identifying user champions, meticulously reviewing both acceptable and less effective responses, and pinpointing areas where the structure of documentation and the wording of prompts required better alignment. "That is the contribution of a domain expert: knowing what ‘good’ looks like in practice," Tay articulated.
The Future of AI Adoption: Synergy of Expertise
Tay firmly believes that the organizations poised for scaled success in AI adoption will be those that effectively harmonize deep technological expertise with robust domain ownership. This synergy is essential for generating not just answers, but outputs that are demonstrably trusted and directly actionable for informed decision-making.
The implications of this approach extend beyond individual organizations. As AI becomes more deeply embedded in business processes across Asia and globally, the emphasis on human oversight and accountability will become increasingly paramount. Regulatory bodies and industry standards are likely to evolve to reflect this reality, demanding clear lines of responsibility for AI-driven outcomes. Furthermore, the continuous development of AI literacy among the workforce, coupled with a strategic focus on change management, will be critical determinants of whether organizations can truly harness the transformative potential of artificial intelligence. The pilot at DBS Bank, while a specific case, offers a microcosm of the broader challenges and opportunities inherent in the ongoing AI revolution, highlighting that the true measure of success lies not in the technology deployed, but in the intelligent, responsible, and human-centric way it is integrated and utilized.
