AI TRENDS | Anthropic to Launch Research System for Scientific Experiments

Anthropic is preparing to launch a research system that lets software control scientific and engineering equipment for experiments, its first tool built for work in the physical world. According to Sina Finance, the Claude chatbot developer plans an IPO and could reach a valuation of $2 trillion. The new tool, called Model Hardware Standards, would let AI agents autonomously operate a wide range of programmable devices, including precision microscopes, liquid handlers, lasers, and robotic arms. Anthropic said it has tested the technology with laboratories and hardware makers in biotechnology, robotics, and quantum computing, including the Howard Hughes Medical Institute, Carnegie Mellon University, Genentech, and QuEra. Anthropic scientist Alec Kemeny said he saw Claude observe living brain tissue for the first time in a neuroscience experiment and used it to locate the alveus, a specific brain structure he had not known well before. Anthropic said it plans to open-source the new technology after further optimization to reduce non-forced errors. Anthropic CEO Dario Amodei said society understands the risks of artificial intelligence in biology less well than risks in areas such as cybersecurity. Anthropic and other AI developers have warned that malicious users could use large models to create new pathogens or biological weapons. In June, Anthropic launched Claude Science for researchers, with uses including rendering protein 3D structures and drug development. Artificial intelligence is becoming increasingly important in drug development at major pharmaceutical companies, including Eli Lilly, which has partnered with Nvidia and invested earlier this year in AI drug developer Insilico Medicine. Meanwhile, Sir Demis Hassabis, founder of Google DeepMind, stepped down from day-to-day management to focus on Isomorphic Labs, the AI drug discovery startup he founded.