Wang Tianmiao's Three Questions to Embodied AI Leaders: Large Models Are Still on "Dial-up Internet" – How Will Robots Secure Real Orders? | WRC 2026

At the roundtable forum of the 2026 World Robot Conference, Beihang University Professor Wang Tianmiao raised sharp questions regarding the current state of the embodied AI industry. He compared current large model capabilities to "dial-up internet," questioning whether they possess the ability to solve complex problems in the physical world, and how robots can obtain real commercial orders.

Over the past two years, large model technology has driven the hype around embodied AI, with various demo videos attracting attention. However, the industry is facing pressure to transition from technical demonstrations to actual deployment. As capital cools down, companies must prove the reliability and economics of their products in real production lines, rather than remaining solely in laboratory environments.

This question reveals the core challenge facing the scaling of embodied AI. The progress of large models in general cognition needs to be deeply integrated with robots' physical control and environment perception. The industry needs to break through bottlenecks in computing power, algorithms, and hardware costs to achieve the leap from "demonstrable" to "functional," promoting the true realization of AGI in the physical world.

This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.