Anthropic has announced a new framework aimed at enabling artificial intelligence agents to safely navigate and interact with various physical systems, including microscopes, liquid-handling equipment, quantum computing hardware, manufacturing machines, and robotic arms. The framework, named Model Hardware Standard, outlines rules for how AI agents should interact with hardware, reflecting the belief that AI can significantly enhance scientific research and manufacturing if deployed safely.
The company plans to collaborate with trusted partners to ensure safety measures are in place before the framework is widely implemented. While there are concerns about potential misuse, such as the development of biological weapons, Anthropic asserts that built-in guardrails within AI models should mitigate risks of malicious use.
Alek Kemeny, a quantum physicist at Anthropic, emphasized the goal of accelerating scientific progress, stating, "How do we close the loop between accelerating literature review and data analysis—and bring that power to the experimental world?" AI agents, which are seen as the next evolution beyond chatbots, are designed to perform tasks and could potentially operate various hardware.
Several startups, including Periodic Labs, LILA Sciences, Edison Scientific, and Discovery Loop, are also exploring AI-driven scientific discovery. These companies aim to automate the development and testing of scientific hypotheses, creating a recursive loop that enhances the scientific discovery process.
Jonah Cool, an experimental biologist at Anthropic, noted that configuring scientific equipment typically requires specialized knowledge. AI could simplify this by automating the engineering processes needed for machines to communicate effectively.
Anthropic is currently working with manufacturers to refine the standard. Kemeny mentioned that the new framework could optimize the behavior of multiple robotic systems on factory lines without the need for custom code.
Despite recent controversies surrounding AI agents, including instances where they have hacked into external systems, Anthropic believes that the new standard will help mitigate risks associated with AI's interaction with physical systems. The company previously introduced the Model Context Protocol, which governs how AI models engage with software programs.