IROS 2025 Debate: Humanoids Will Soon Replace Most Human Workers: True or False

In this talk

1. 3:18

   The case against near-term replacement

   Hai Zhao of Tsinghua University argues that reliability, adaptability, safety, and maintenance costs make humanoids unready to replace workers soon. She also warns that the words “soon” and “replace” can fuel fear and impatience, and says robots should have time to assist people and develop gradually.

2. 12:53

   Physical technology takes time to spread

   Andrea Kier, president of Silicon Valley Robotics, compares humanoids with automobiles, which took decades to reach mass production. She notes that about 3 million industrial robots and 50 million robot vacuums have been deployed, and argues that producing billions of humanoids would take time and could create jobs. She also points to the difficulty of real-world work, where objects and environments are messy, heavy, and unpredictable.

3. 29:43

   A 2050 vision, with hard problems to solve

   Tamim argues that humanoids could replace many workers by 2050, citing advances in mechatronics and AI, labor shortages, and the fit of humanoid bodies with existing infrastructure. He says progress depends on robust hardware, manipulation, safety, trust, and embodied intelligence. Other panelists favor assistive roles, noting challenges such as grasping cables, assembling tiny parts, and providing the empathy patients expect from surgeons.

4. 40:53

   Where robots may help, and what remains human

   In discussion, panelists debate speed, actuators, and whether robot production may create jobs. Near-term possibilities include factory night shifts and machine tending, while sewing and care work are described as difficult tasks. Examples from cooking and surgery highlight a distinction between handling physical tasks and providing creativity, judgment, or empathy. The host reports that support for the proposition fell from roughly 20% to 10% in the audience vote.