ChatGPT develops its first robot which can harvest tomatoes

The team followed ChatGPT’s design decisions to the T

“Even though Chat-GPT is a language model and its code generation is text-based, it provided significant insights and intuition for physical design, and showed great potential as a sounding board to stimulate human creativity,” said Josie Hughes, head of the Computational Robot Design & Fabrication Lab at EPFL’s School of Engineering, in a press release.

In the ‘ideation’ phase, the team asked ChatGPT questions like ‘What are the greatest future challenges for humanity?’ to outline the robot’s purpose, design parameters, and specifications. Choosing global food supply as their challenge, in their discussions with the chatbot, they collectively came up with the idea of creating a tomato-picking robot.

The team then asked ChatGPT ‘What features should a robot harvester have?’ For this, the researchers relied on the chatbot’s access to global data from academic publications, technical manuals, books, and media to provide answers to their various prompts.

After they set on a tomato harvesting robot, the researchers asked ChatGPT more specific questions like ‘What shape should the gripper have?’, and asked the LLM to make technical suggestions including materials and computer code for controlling the device, they explained.

The robotic community must identify how to cultivate these AI tools

During this phase, ChatGPT came up with useful suggestions like recommending to use of a gripper made of silicon or rubber to avoid crushing the tomatoes and suggesting to use of a dynamixel motor to drive the robot.

“While computation has been largely used to assist engineers with technical implementation, for the first time, an AI system can ideate new systems, thus automating high-level cognitive tasks. This could involve a shift of human roles to more technical ones,” said Francesco Stella, lead author of the study and a Ph.D. student at EPFL.

One cannot ignore issues like bias, plagiarism, and hallucination when it comes to dealing with LLMs like GPT. The researchers caution that the role of LLMs must be carefully evaluated, as it’s unclear if a design generated by an LLM like ChatGPT can be considered novel.

“In our study, ChatGPT identified tomatoes as the crop ‘most worth’ pursuing a robotic harvester. However, this may be biased towards crops that are more covered in literature, as opposed to those where there is truly a real need. When decisions are made outside the scope of knowledge of the engineer, this can lead to significant ethical, engineering, or factual errors,” said Hughes.

Hughes also noted that the robotic community must identify how to cultivate these AI tools for the advancement of robots in a sustainable, socially empowering, and ethical way.