“Thinking AI is Coming”… The AI Landscape Changed by Agent Workflow
Stanford University Professor Andrew Ng: “Korean Startup Ecosystem Will Become Silicon Valley’s Key Partner”
“Just as electricity led to the industrial revolution, AI will become a general-purpose technology that permeates all industries.”
Stanford University's Andrew Ng, a world-renowned AI expert, presented a vision for the future of AI innovation at the UKF 2025 Startup Summit. Professor Ng, who is well-known as the co-founder of Google Brain and the founder of AI education platform Coursera, said, "Many people still underestimate the diversity of AI applications," and predicted the future direction of AI development after ChatGPT.
Professor Eung cited 'agent workflow' as the most notable AI trend. While existing AI was limited to simple question-and-answer, agent workflow enables AI to perform repetitive tasks such as thinking, researching, and revising results. "Just as humans go through the process of creating an outline, finding materials, writing a draft, and then revising when writing, AI is now capable of more systematic and in-depth work," he explained.
He also highlighted a new change in the AI technology stack. Above the semiconductor layer, the cloud layer, and the basic model layer, a new “agent orchestration” layer has emerged. This is a software layer that effectively combines multiple AI models to perform more complex tasks.
In particular, Professor Eung expressed his expectations for the Korean startup ecosystem. “Korea has global competitiveness in the semiconductor sector, which is essential for AI innovation,” he said, adding, “Collaboration between Silicon Valley and Korean startups will become more important.”
He also predicted a revolutionary change in the AI development process. With the advancement of generative AI, the cost and time for prototyping have been greatly reduced, allowing companies to attempt more experiments and innovations. “AI system development that used to take 6-12 months is now possible in just 10 days,” he said, emphasizing that “it is important to move quickly but be responsible.”
Professor Eung suggested the following as future directions for AI innovation: ▲an image processing revolution beyond text processing ▲creation of new opportunities due to reduced data gravity ▲increased importance of unstructured data processing. He predicted, “Just as text analysis has created great economic value, image analysis will open up new opportunities in the future.”
He also shared his experience in fostering startups through AI funds. “We are working with corporate partners to create new AI startups,” he said. “We especially hope to see more innovative startups emerge through collaboration with Korean companies.”
Finally, Professor Eung emphasized the importance of AI education. “The fastest-growing course on Coursera in 2023 was ‘Generative AI for Everyone,’” he said. “It is important to understand and utilize the business opportunities of AI even if you do not have technical knowledge,” he advised.