HKUST's Justin Long: The Agent Moat Has Not Yet Solidified, with Model Differences More in Efficiency Rather Than Disruptive Breakthrough

BlockBeats News, April 21st: During the "Decoding Web 4.0: When AI Agent Takes Over On-Chain Authority" roundtable discussion, Kelvin Tan, Associate Vice President of Hong Kong University of Science and Technology, discussed the "Agent Moat," emphasizing that the different AI Agents rely on distinct model training paths and technical systems, resulting in a significant variance in practical user experience. Recently, some new models and tools have shown superior performance in generation quality and execution efficiency, and have even demonstrated a higher ceiling in development output.

However, he pointed out that at the current stage, this difference has not yet formed a decisive delta, leaning more towards "efficiency improvement" rather than "paradigm breakthrough." In other words, the competition between Agents is still in a rapid evolutionary phase, with no stable and insurmountable technological barrier in sight.

The current AI Agents and large models iterate at an extremely rapid pace, with new products or capabilities emerging almost every week, driving continuous industry advancement. However, from the perspective of practical usage and business decision-making, the need for continual high-frequency tracking of these changes still requires careful evaluation.