The idea of humanoid robots taking over everyday tasks has long been a captivating vision, but it remains a distant dream, despite recent breakthroughs in artificial intelligence (AI). Robots like Pepper, introduced in 2014, sparked a wave of excitement but ultimately ended in disappointment. While the hype was real, the practical applications of humanoid robots have not yet materialized as expected.

Pepper, a humanoid robot powered by AI, was heralded as a machine that could autonomously interact with humans and perform various tasks. Backed by major investors like SoftBank, Alibaba, and Foxconn, the vision was ambitious: to make robots a ubiquitous part of our daily lives. However, production was halted in 2021 after only 27,000 units were sold, leaving Pepper a symbol of unfulfilled potential.

Despite the setback, the dream of robots that can perform tasks like cooking and cleaning is still alive, driven by the recent surge in AI advancements. Companies like Nvidia, the world’s second-largest company by market capitalization, are betting big on AI-powered robots. Nvidia’s CEO, Jensen Huang, believes that “the next wave of AI is physical AI” — technology capable of understanding the laws of physics and working seamlessly alongside humans.

This new AI-driven approach aims to tackle the massive challenge of teaching robots to understand and interact with the physical world. By applying AI techniques previously used for tasks like predicting protein folding or generating realistic text, robotics startups are striving to create machines that can manipulate objects, understand environments, and perform tasks that once seemed reserved for human hands.

However, as exciting as the vision sounds, the reality of creating a robot that can perform complex household chores is far more complicated than just improving AI algorithms. The physical challenges remain enormous and are the primary obstacle to widespread adoption.

For example, human limbs move via muscles, while robots rely on motors for movement. Each motor requires additional components such as gears and transmissions, which add bulk, cost, and potential points of failure. Even if we solve the movement problem, robots face the challenge of sensing their surroundings. A human hand can feel the softness of fruit or the heat of a pan, while a robot relies on imperfect substitutes like machine vision and AI to gather information about the physical world.

Powering these robots is another significant hurdle. While factory robots are tethered to power sources, humanoid robots will likely require batteries, which come with trade-offs in terms of size, power, and cost. These are just some of the many physical limitations that AI cannot magically solve.

So, what role does AI play in the future of robotics? Rather than focusing on creating entirely new robots, AI is more likely to enhance the capabilities of existing machines. A prime example is self-driving vehicles, where AI can improve the way cars navigate the world, but the physical infrastructure remains the same. In the realm of robotics, AI-powered advancements will make existing devices, such as industrial robotic arms and robot vacuum cleaners, more efficient, versatile, and safer to use.

In fact, the most practical applications for AI in robotics are in areas where robots are already functioning in controlled environments. For instance, in China, robot waiters deliver food to hotel guests, and AI-enhanced robotic arms are already used in factories to perform precise tasks.

While AI-powered robots will undoubtedly continue to evolve, the vision of a humanoid robot capable of performing household chores remains a far-off dream. For now, robots like Pepper, once imagined as the future of everyday life, continue to serve as a reminder of the gap between innovation and practical reality. The future of AI in robotics may not bring us androids just yet, but it will undoubtedly change the way we interact with technology — in ways both big and small.