The Rise of the Machines: How Robots Are Getting Smarter

Can you put the nuts in the lunchbox?” Slowly but surely, a humanoid robot developed by Apptronik picks up a small box and places it inside a lunchbox. “Can you put the orange in there too?” From a small plate with different pieces of fruit, the robot slowly but surely picks up an orange again and places it in the lunchbox. Wouldn’t it be great to have a robot at home that could prepare your lunchbox every day? And help you with other routine tasks, such as hanging out the laundry, preparing meals or simply keeping the house tidy? And what if, in addition to helping around the house, it could also help you at work?

The example described here is part of one of the latest demonstration videos of Gemini Robotics, an Artificial Intelligence (AI) model developed by Google specifically for robots. In recent years, digital assistants have evolved significantly thanks to the development of so-called large language models (LLM), which allow tools such as ChatGPT, Gemini or Claude to understand and write text with a quality very close to that of a human. But what began as a text-based interaction quickly evolved into so-called multimodal models – systems that understand text, images, sounds, videos, programming or even different languages… In a way, they are becoming more comprehensive, more generalist intelligence.