Advances in artificial intelligence (AI) offer tantalizing glimpses of a future where robots mimic human movement, but the path there is fraught with challenges. Elon Musk’s Optimus, for instance, is touted to revolutionise labour with human-like dexterity, but researchers remain sceptical. The real question is whether the same AI techniques that power chatbots can handle the complex physical world—or if a whole new approach is needed.

The debate centres on whether humanoid robots are the key to a future filled with generalist machines, or if simpler, more practical designs will suffice. Google’s Gemini Robotics, for example, shows that simpler, task-specific robots are already making progress. Yet, the march towards fully autonomous, versatile robots remains slow and uncertain.

Roboticists believe that a generative AI revolution could endow machines with the physical intuition needed to match human capabilities, but the journey is far from smooth. The challenge lies in developing robot policies that can enable machines to understand, navigate, and interact with the physical world in a fluid and intuitive manner. Until then, humanoid robots may remain a distant dream.

The hype over timelines often obscures the painstaking progress being made in robotics labs. While Optimus might not be bringing home the popcorn anytime soon, the advancements in generalist AI are pushing the boundaries of what machines can achieve. The question remains: will current AI methods be enough, or do we need a new path?