Chen Gen: ChatGPT cooperates with humans to develop robots

文/Chen Gen

The emergence of ChatGPT has greatly promoted the development of the field of artificial intelligence, and thus triggered earth-shaking changes in the field of intelligent technology. Today, AI can not only process and understand large amounts of text data, and use this data information to communicate with humans, it can even collaborate with humans to develop and design robots, which makes people wonder, "What else is there?" What ChatGPT can’t do”!

According to the latest report in the magazine "Nature Machine Intelligence", on the road to exploring the cooperation between humans and large language models (LLM), scientists from the University of Delft in the Netherlands and the Ecole Polytechnique Fédérale de Lausanne (EPFL) in Switzerland used ChatGPT Designed and developed a tomato harvesting robot.

"Although ChatGPT is a language model and its code generation is text-based, it provides important insights and intuition for physical design and shows great potential as a sounding board for inspiring human creativity." Engineering Josie Hughes, director of the college's Computational Robot Design and Manufacturing Laboratory, said, "This may change the way humans design robots, while enriching and simplifying the process."

Throughout the collaborative research and development process, researchers asked ChatGPT questions to guide them in the direction selection, creative conception and detailed design of robot development, including which building materials will be used and the creation of computer code to control it. Currently, LLM cannot generate entire computer-aided design models, evaluate code, or automatically manufacture robots. But with technical advice provided by ChatGPT, the researchers built their robotic gripper and tested it in the real world, using it to pick tomatoes, with success.

"While computers are mainly used to help engineers with technology implementation, artificial intelligence systems can for the first time conceive new systems that automate high-level cognitive tasks." EPFL doctoral student Francesco Stella said, "This may involve transforming human roles into more A technical role."

The successful development of the tomato harvesting robot demonstrates the great potential of collaboration between human scientists and LLM to transform the design process. However, researchers raise different assumptions and corresponding ethical and common sense risks regarding the manner and depth of their collaboration. If humans take artificial intelligence as an "inventor" to the extreme and blindly apply the robot design input it provides, it is obviously undesirable and involves huge risks. But retaining humans as inventors and using AI to troubleshoot, debug, and handle tedious or time-consuming projects to refine the design process is an upgrade over human invention.

There are also problems and risks of plagiarism and intellectual property rights when using artificial intelligence for R&D and design. Additionally, it is unclear whether the design generated by LLM can be considered "novel" because it uses existing knowledge. "In our study, ChatGPT identified tomatoes as the crop 'most worthy' of being pursued by robotic harvesters, which may be biased towards crops more covered in the literature rather than those that are truly in need," the researchers said. When decisions are made beyond an engineer's knowledge, they can result in significant ethical, engineering, or factual errors."

The potential for humans and artificial intelligence to work together is undoubtedly huge. However, humans must have a clear sense of boundaries and a clear sense of how to use these powerful tools to accelerate the development of robots in an ethical, sustainable and socially empowering way. Based on this, "We believe that LLM will open up many exciting possibilities and become a force for good." The researchers said.

The above is the detailed content of Chen Gen: ChatGPT cooperates with humans to develop robots. For more information, please follow other related articles on the PHP Chinese website!