Anthropic has announced the introduction of MHS, a new hardware standard designed to enable AI agents to control and interact with the physical world. The standard allows models to sequence steps across instruments, adjust parameters, and operate physical equipment with greater ease and efficiency. For instance, a model like Claude can adjust a laser, check the results via a separate camera, and then repeat the process to automatically calibrate the whole system.
MHS could also enable an AI model to focus a microscope, analyze the results, decide what part needs more observation, and then automatically move the microscope to the relevant section to continue the experiment. In a video, Anthropic demonstrated Claude reasoning how to get a robotic arm to pick up an aluminum can, even though it had not been specifically trained on the required steps.
Anthropic says MHS includes a standardized tagging system to describe hardware's real-world constraints for models that may have been trained more in the virtual world. These tags can be integrated into a reference file that provides an AI model with crucial information about a device it has no previous training experience with. The company is working with a first group of scientific research labs and advanced manufacturers during an MHS preview period, including Amazon Web Services, Hugging Face, Raspberry Pi, Automata, and Universal Robots.
These partners will help Anthropic build safety evaluations and develop best practices for AI systems operating physical equipment. After the preview period, the plan is for MHS to become an open-source and agent-agnostic standard for integrating AI and physical systems. In early testing with scientific partners over the past year, Anthropic saw MHS reduce the time it took to integrate devices, making it possible to iterate faster in a variety of experimental settings.
According to Anthropic, this could lead to significant advancements in various fields, as faster hypothesis testing enables the creation of general technologies at a rapid pace. As Kemeny noted in a promo video, "If you can test hypotheses faster, you could create general technologies faster. This is how a century of progress can condense into a decade." The potential applications of MHS are vast, and its development could mark a significant milestone in the integration of AI and physical systems.
Este artículo fue escrito con la asistencia de IA.
News Factory APP - noticias agénticas para impulsar tu SEO y AEO.