When a passenger's rescheduled flight needs adjustment, a food delivery arrives with a broken item, or an online purchase goes missing, customers expect swift resolution. Yet increasingly across Malaysia and the region, these straightforward service failures become exercises in digital frustration. The common thread? Companies replacing human customer service representatives with artificial intelligence-powered chatbots designed primarily to reduce costs rather than resolve problems. As these systems proliferate across airlines, e-commerce platforms, and delivery services, consumer complaints have spiked, revealing a deeper flaw in how Malaysian businesses are implementing automation technology.
The Malaysia Cyber Consumer Association has documented a significant rise in complaints about automated customer support systems, according to its president Siraj Jalil. The primary grievance centres on what insiders call the "infinite loop" phenomenon—a deliberate or accidental design flaw where chatbots remain hard-coded to recognise only specific keywords. When a customer presents a problem that doesn't fit the system's narrow parameters, the bot endlessly cycles through the same FAQ links, offering no escape route to human assistance. This creates what frustrated users on social media platforms like X and Reddit describe as an exhausting, circular experience that treats their time with contempt.
The root cause, according to technology strategists, lies in how Malaysian companies measure chatbot success. Henrick Choo, managing director of NTT Data Malaysia, identifies a perverse metric that has infected corporate decision-making: companies celebrate keeping customers away from human agents rather than celebrating actual problem resolution. Under cost pressure, many organisations have adopted chatbots as a volume-reduction tool first and a solution-delivery tool second, if at all. This inverted priority creates an immediate consumer backlash. "Customers sense this immediately," Choo observes. "They feel the bot is there to block them rather than help them." The irony, he notes, is that this approach often backfires spectacularly—more frustrated customers mean more repeat contacts, escalating complaints, and reputational damage that far exceeds the savings from reduced agent headcount.
Academic research from Johns Hopkins University provides empirical weight to these observations. A comprehensive study on AI chatbots in customer service identified this phenomenon as "gatekeeper aversion"—the psychological resistance consumers develop toward automated systems perceived as barriers rather than helpers. Associate Professor Evgeny Kagan and his research team found this aversion remarkably persistent. Users enter interactions with chatbots already expecting failure, and they actively resist engagement. This negative perception becomes even more pronounced when the system offers no obvious path to human escalation, leaving customers stranded in digital limbo, unsure whether persistence will eventually yield help or merely prolong their ordeal.
The frustration intensifies dramatically when chatbots finally surrender and route users to a human agent. Siraj explains the second layer of dysfunction: "contextual blindness." Many systems completely discard conversation history if a connection refreshes, times out, or transfers between channels. When customers finally reach a live representative, they encounter cheerful automated greetings asking "How can I help you today?" as if the entire prior interaction never occurred. Customers must then recount their full grievance from scratch, a repetitive process that consumers universally describe as disrespectful and draining. If the live chat disconnects, the cycle begins anew—back into the queue, back to square one, back to explaining everything again.
Choo identifies the handoff as the critical failure point where companies lose customer trust irreversibly. He argues that customers don't inherently resist self-service; they become trapped when exiting what he terms the "doom loop"—those cycles of repeated prompts and failed resolutions—requires increasingly desperate tactics to reach a human. The solution lies in context preservation. When a customer has explained their issue to an AI system, the human agent should inherit the complete conversation transcript, customer profile history, previous transaction records, emotional sentiment, and recommended next steps. Context, Choo emphasises, represents the boundary between efficiency and frustration.
Yet this contextual integration requires far deeper systemic work than most Malaysian companies have invested. The problem extends beyond the chatbot's conversational capabilities into the underlying technical architecture. Data integration, escalation rules, permission structures, and tool access constitute invisible but critical failures in experience design. Many companies connect chatbots to knowledge bases while failing to connect them to the actual systems where work occurs—customer relationship management platforms, billing systems, identity verification services, approval workflows, compliance databases. A chatbot can retrieve FAQ answers effortlessly, but resolving an account issue demands access to the same CRM, billing, and authorization tools that human agents wield. Without this integration depth, the chatbot becomes a sophisticated knowledge retriever incapable of taking actual action, leaving customers in a limbo between information and resolution.
Another widespread flaw stems from the assumption that companies can simply feed all their documentation into a large language model and expect perfect operation. Khalil Nooh, CEO of local language model firm Mesolitica, warns of what he calls "knowledge-base rot"—a common condition where legacy documentation contains obsolete pricing, conflicting policies, expired terms, and contradictory information. When AI systems ingest such corrupted data sources, their retrieval precision collapses and the models begin "hallucinating," generating plausible-sounding but factually incorrect responses. Malaysian companies often haven't prepared their internal knowledge repositories for AI consumption, leaving the technology to extract signal from noise and deliver inaccurate guidance to customers.
Fundamentally, Nooh and other experts identify a strategic misconception permeating Malaysian corporate deployments: the belief that AI chatbots should progressively assume entire customer support functions, with human agents relegated to a peripheral role. This approach ignores the practical reality that sophisticated problems require human judgment, contextual understanding, and authority to act. When companies eliminate experienced human frontline agents while implementing chatbots without proper escalation pathways or supporting systems, they don't eliminate customer service—they fragment it. The customer experiences a disconnected sequence of failed automation attempts before potentially reaching an overwhelmed human agent lacking institutional knowledge or troubleshooting context.
For Malaysian consumers and businesses alike, the implications are significant. Customers increasingly distrust AI-mediated support channels, viewing them as obstacles to genuine help rather than efficiency improvements. Businesses, meanwhile, generate reputational damage and customer churn that outweighs cost savings. Competitors who implement chatbots as genuine first-contact resolution tools—backed by proper system integration, contextual handoffs, and appropriate human escalation—will capture market share from companies treating automation as a cost-cutting sledgehammer. The technology itself remains valuable, but only when deployed with authentic customer-first design rather than agent-reduction metrics.
The path forward requires Malaysian organisations to fundamentally reframe their approach. Rather than asking "How many customers can we keep away from agents?", companies should ask "How many issues can we resolve on first contact?" This reorientation demands investment in system integration, knowledge base quality, agent empowerment, and transparent escalation pathways. The chatbot's role should shift from gatekeeper to efficient first-responder, capable of gathering context and either solving problems directly or smoothly transferring informed customers to capable humans. Without these changes, Malaysian consumers will continue encountering the doom loops that define current experience—and competitors who get it right will be rewarded accordingly.
