The United Nations’ AI for Good Global Summit 2023 in Geneva brought together visionaries, UN agencies, and investors to explore how artificial intelligence and robotics can accelerate progress toward the 17 Sustainable Development Goals (SDGs) outlined in the 2030 Agenda. More than 50 robots were showcased, including nine humanoid designs aimed at improving health, education, and reducing inequalities. Among them, Grace, Sophia, and Ameca demonstrated capabilities that extend beyond novelty, integrating AI to deliver tangible societal benefits.
Recent years have seen a marked leap in humanoid robotics performance. Boston Dynamics’ Atlas, a bipedal robot, has exhibited dynamic mobility—running, jumping, and climbing with precision. This agility represents a significant departure from earlier humanoid designs, which often struggled with stability and terrain navigation. Engineered Arts’ Ameca, powered by generative AI, exemplifies advancements in human-robot interaction. Capable of composing poetry or sketching on demand, Ameca’s responsiveness underscores the integration of creative and conversational AI into mechanical platforms.
These improvements over prior generations are substantial. Mobility systems have evolved through advanced actuation, real-time perception, and model-predictive control, enabling robots like Atlas to perform tasks once considered impractical for machines. Interaction capabilities have been transformed by AI models that interpret language, detect emotion, and respond with nuanced expressions, as seen in Ameca’s lifelike facial mechanics.
Research leadership in this domain spans institutions and companies globally. Boston Dynamics continues to refine high-mobility robots such as Atlas and Spot. Engineered Arts develops expressive humanoids for interaction-rich environments. Carnegie Mellon University and MIT’s CSAIL contribute foundational research in robotics and AI algorithms, while Stanford’s AI Lab advances natural human-robot communication.
The AI for Good initiative, central to the Summit, promotes AI systems that are transparent, explainable, and accountable, with respect for privacy and ethics. Its mission is to ensure AI benefits humanity, tackling challenges from poverty eradication to quality education. Collaboration across nations, academia, industry, and civil society is encouraged to maximize AI’s positive impact while addressing its societal implications.
Notable humanoids illustrate the breadth of application. Grace, from Hanson Robotics, is tailored for healthcare, capable of taking vital signs and assisting patient communication. Sophia, also from Hanson Robotics, blends AI, visual processing, and facial recognition to conduct conversations and adapt to human behavior, earning recognition as the UNDP’s first non-human Innovation Champion. Ameca serves as a development platform for AI and machine learning, with advanced facial expression simulation.
The industrial ecosystem for AI-powered humanoids includes UBTech Robotics, whose Walker X navigates autonomously and performs service tasks; Macco Robotics, specializing in hospitality-focused machines; Promobot, delivering customer service and security solutions; Tesla, applying its autopilot AI to humanoid form factors; and Shadow Robot Company, designing dexterous manipulators for sensitive or hazardous work.
Humanoids are deployed across diverse sectors. In healthcare, they assist with rehabilitation, patient monitoring, and elder care. In entertainment, animatronic and stunt-capable robots enhance theme park experiences. Procedural vocations benefit from humanoids’ ability to operate human-designed tools and equipment. Cognitive science and biomechanics research leverage humanoids to model human motion and learning processes.
Several designs have gained global recognition. Alter 3, from Osaka University and Mixi, integrates musical performance capabilities. Armar-6, from Karlsruhe Institute of Technology, supports industrial collaboration. Astro, by Apptronik, applies machine learning to logistics tasks. Digit, from Agility Robotics, handles package movement with adaptive balance control. Beomni, from Beyond Imagination, combines remote teleoperation with AI learning. JiaJia, developed in China, reflects advances in lifelike appearance. Walker X demonstrates service versatility, while Atlas continues to push the limits of robotic agility.
Realism in humanoid robots has advanced through precise actuation and material science. PediaRoid, developed in Japan, can simulate physiological and emotional states, while Nikola conveys basic emotions via facial muscle actuation. Engineered Arts’ Adran head features 22 actuators for detailed eye and mouth movement. Hanson Robotics’ Grace simulates over 48 facial muscles, and Promobot’s RoboC delivers hundreds of expressions through 16 moving facial components.
Boston Dynamics remains a benchmark in robotic mobility engineering, with platforms like Atlas embodying the synthesis of mechanical design, control theory, and AI perception systems.