By: Luu Quy (VnExpress.net) | Published: May 2026
While Artificial Intelligence is increasingly capable of executing highly complex tasks, its growing power paradoxically demands greater human supervision and operational expertise. This was the core message delivered by Dr. Ed H. Chi, Research Vice President at Google DeepMind, during a recent panel discussion.
At the forum “AI: Shaping Human-Centered Innovation” held on May 27 in Hanoi, leading global experts gathered to discuss how AI is fundamentally restructuring education, workforce dynamics, and the nature of innovation. However, the consensus among panelists was clear: AI is not a plug-and-play technology. Seamlessly integrating AI into an organization requires an elite workforce of engineers and specialists capable of controlling, auditing, and optimizing these advanced systems.
“Much like the advent of industrial machinery, you cannot simply purchase hardware, place it on the factory floor, and expect it to generate value automatically,” Dr. Ed H. Chi explained. “You need highly skilled engineers to operate and optimize them. The exact same principle applies to AI.”
The High Cost of Autonomous AI Failures
While AI can drastically improve efficiency in software engineering, data analytics, and information processing, these systems are far from infallible. Referencing internal research from Google, Dr. Chi noted that while AI models generally perform exceptionally well, their failures can be uniquely catastrophic.
The Illusion of Perfection: “Hallucinations”
Large Language Models (LLMs) remain highly susceptible to “hallucinations”—a phenomenon where the system generates false data or deeply flawed logic with immense confidence. Because of this, enterprises must rigorously mitigate these errors before deploying AI systems at scale. Total automation is a myth; humans must remain in the loop at critical decision-making junctions.
To illustrate, Dr. Chi compared the evolution of AI to the automotive industry:
- The Learning Curve: When cars were first invented, humanity had to learn the mechanics of driving and traffic management.
- The Autopilot Dilemma: Even today, with the existence of autonomous vehicles, relying completely on software to navigate chaotic, real-world traffic remains an unsolved challenge.
- “We cannot simply buy an AI solution from Google, OpenAI, or any other developer, drop it into an enterprise, and expect it to run seamlessly on its own,” he emphasized.
Sharing a real-world example from the frontier of advanced science, Dr. Luong Minh Thang, a Research Director at Google DeepMind who leads world-class teams developing AI for complex mathematics, recalled an incident involving theorem solving:
“The AI produced a solution that looked incredibly elegant and plausible at first glance. However, after several hours of rigorous manual verification, our team discovered that the AI’s conclusion was entirely wrong.”
Dr. Thang pointed out that while modern users are constantly bombarded with AI-generated suggestions, they must develop the technical acumen to evaluate whether those outputs are accurate, safe, and aligned with their specific objectives.
A Human-Centered Approach to Healthcare
In the medical sector, the implementation of AI requires even greater caution and a deeply human-centric philosophy, according to Jean Desombre, Founder and Partner at Pacific Gateway Partners. While AI holds massive potential for drug discovery, identifying novel therapies, and parsing biomedical big data, she insists that the core purpose of the technology must be questioned first.
- Will this AI genuinely improve clinical outcomes for the patient?
- Will it alleviate administrative burdens so doctors and nurses can spend more quality time at the bedside?
- Will it measurably elevate the patient’s quality of life?
“These are the absolute baseline questions that must be answered before a single line of code is deployed in healthcare,” Desombre stated.
Education in the Age of Artificial Intelligence
Faced with these realities, how must education evolve? Dr. Luong Minh Thang believes that the ultimate goal of modern education should be fostering self-determinism and critical thinking. “We must build a mindset of healthy skepticism. Never blindly accept an AI’s first recommendation,” he advised. “Maintain relentless curiosity, high ambition, and a sharp critical lens.”
Professor Po-Shen Loh, a renowned mathematician from Carnegie Mellon University (USA), echoed this sentiment, arguing that the rise of AI forces education to return to its most fundamental question: Why do humans learn?
OLD PARADIGM NEW PARADIGM
┌─────────────────────┐ ┌─────────────────────┐
│ Learning to Know │ ──> │ Learning to Think, │
│ │ │ Understand & Act │
└─────────────────────┘ └─────────────────────┘
According to Professor Loh, when AI can retrieve information faster than any human brain, the educator’s role shifts from delivering raw data to cultivating clear thinking, effective communication, and the capacity to generate unique societal value.
Drawing from his extensive experience coaching elite math students, he noted that the true challenge in modern education is not a lack of content, but inspiring a genuine desire to learn. Effective education must treat students like “valued clients”—understanding what captivates them and maintains their long-term drive. A young person who can think logically, ask profound questions, and connect empathetically with others will always outcompete a machine when solving complex societal issues.
Vietnam’s Multi-Million Dollar Pivot
From a macroeconomic perspective, Truong Gia Binh, Chairman of FPT Corporation, views this AI wave as a historic inflection point for both enterprises and developing nations. He believes Vietnam is uniquely positioned to become a global hub for AI transformation services.
In the near future, the strength of an organization will not be measured merely by its headcount, but by its ability to orchestrate hybrid ecosystems where humans and AI agents work side-by-side. If democratized correctly, AI can exponentially boost labor productivity and sharpen national competitiveness. However, achieving this requires absolute growth determination, modernized education, and the Vietnamese spirit of rapid learning and bold execution.
To catalyze this vision, Binh announced the establishment of the Au Lac AI Alliance alongside the launch of the Au Lac Grand Prize—a $1 million USD initiative designed to challenge young local talents to tackle massive global problems, master human-AI collaboration, and forge a new competitive edge for the nation in the age of intelligence.
Summary of Key Event Credentials
- Forum: AI: Shaping Human-Centered Innovation
- Date: May 27, 2026
- Location: Hanoi, Vietnam
- Key Speakers: Dr. Ed H. Chi (Google DeepMind), Dr. Luong Minh Thang (Google DeepMind), Prof. Po-Shen Loh (Carnegie Mellon University), Jean Desombre (Pacific Gateway Partners), Truong Gia Binh (FPT).