ChatGPT, the advanced language model developed by OpenAI, has been utilized to design a cutting-edge robot.

 Is there anything the open AI platform ChatGPT can't handle? Researchers at TU Delft and EPFL have explored this question further and investigated whether ChatGPT can design a robot. Their findings, published in Nature Machine Intelligence, discuss the benefits and risks associated with this collaboration.

Cosimo Della Santina, Francesco Stella, and Josie Hughes engaged in a conversation with ChatGPT to tackle the challenge of food supply. They specifically focused on designing a tomato-harvesting robot. ChatGPT provided valuable input during the conceptual phase, expanding the designers' knowledge and suggesting economically valuable crops for automation. During the implementation phase, ChatGPT offered practical recommendations, such as using silicone or rubber for the gripper and employing a Dynamixel motor to drive the robot. This collaboration resulted in the creation of a robotic arm capable of efficiently harvesting tomatoes.

While the researchers found the design process with ChatGPT to be positive and enriching, they also noted a shift in their role as engineers towards more technical tasks. In their publication, they explore different degrees of cooperation between humans and Large Language Models (LLMs) like ChatGPT. In an extreme scenario, where AI provides all the input and humans blindly follow it, the LLM acts as the researcher and engineer, while the human assumes the role of a manager responsible for specifying design objectives.

However, the researchers caution against relying solely on LLM output without verification or validation. LLMs generate answers based on probabilities, which can result in misinformation and bias in the robotics field. Working with LLMs raises additional concerns regarding plagiarism, traceability, and intellectual property.

Moving forward, Della Santina, Stella, and Hughes plan to continue their research on robotics using the tomato-harvesting robot. They are also investigating the autonomy of AIs in designing their own bodies. One of the key questions for the future is how LLMs can assist robot developers without limiting creativity and innovation, allowing robotics to meet the challenges of the 21st century.