AI scientists AI News & Updates
Periodic Labs Emerges with $300M to Build Autonomous AI Scientists for Materials Discovery
Periodic Labs, founded by former OpenAI and Google DeepMind researchers, launched with $300 million in seed funding from major tech investors including Bezos, Schmidt, and Nvidia. The startup aims to automate scientific discovery by building autonomous laboratories where AI-controlled robots conduct physical experiments to discover new materials, starting with superconductors. The company seeks to generate fresh training data from real-world experiments as internet-based data sources for AI models become exhausted.
Skynet Chance (+0.04%): Autonomous AI systems conducting unsupervised physical experiments with self-improvement capabilities introduces incremental risks around loss of experimental control and unintended consequences in material synthesis. However, the domain-specific nature and physical constraints of materials science limit immediate existential risk compared to general-purpose AI systems.
Skynet Date (-1 days): The development of autonomous, self-improving AI systems operating in physical environments represents modest acceleration toward more capable and independent AI agents. The narrow focus on materials science and the physical safety constraints of laboratory environments somewhat limit the immediate timeline impact.
AGI Progress (+0.04%): This represents significant progress toward AGI by demonstrating AI systems capable of autonomous hypothesis generation, physical experimentation, and iterative learning across real-world scientific domains. The integration of physical world interaction with self-improvement loops addresses a key limitation in current AI systems that primarily operate in digital environments.
AGI Date (-1 days): The massive $300 million seed funding and assembly of top-tier researchers from OpenAI and DeepMind significantly accelerates development of autonomous AI agents capable of real-world scientific discovery. The explicit goal of generating new training data from physical experiments addresses the data exhaustion problem that currently limits AI progress, potentially accelerating the broader field.