What is Kosmos, the AI system Sam Altman calls 'exciting'?
What's the story
OpenAI chief Sam Altman has praised the development of Kosmos, a next-generation AI scientist from Future House. The system is designed to accelerate scientific discovery and is seen as a major step forward in artificial intelligence (AI) research. Altman called the development "exciting" and predicted that systems like this would have a huge impact on AI in the coming years.
Technological advancement
Kosmos: A major upgrade in AI research
Kosmos is a major improvement over Future House's previous AI scientist, Robin. The new system is now being handled by Edison Scientific, a commercial spin-off that will run the platform while providing free access for academics. Traditional AI models struggle with large amounts of scientific data due to their limited memory capacity. Kosmos tackles this problem by using structured world models, allowing it to leverage information from hundreds of agent runs while focusing on a single research goal.
Operational efficiency
Capabilities and performance
A typical run of Kosmos involves reading around 1,500 scientific papers and executing over 42,000 lines of analysis code. This is far more than what previous systems used in research. Early users estimate one Kosmos run can do work that would normally take a human scientist about six months. Internal testing found almost 80% of Kosmos's conclusions were accurate, demonstrating its potential as a powerful tool for scientific discovery.
Research impact
Discoveries span multiple scientific fields
During testing, Kosmos made seven discoveries in neuroscience, materials science, genetics, and aging research. It confirmed three previously published findings and generated four new insights. These include a suggestion that higher levels of the antioxidant enzyme SOD2 may reduce heart fibrosis in humans and a new molecular explanation for how a genetic variant may lower Type 2 diabetes risk. Edison Scientific is currently validating some of these findings through wet lab experiments.
Accountability
Ensuring traceability and transparency in research
A key feature of Kosmos is traceability. Each conclusion can be traced back to specific lines of code or scientific passages that informed it. This ensures all reports remain auditable and avoids the "black box" problem common in many AI tools. Kosmos is available now at a price of 200 credits per run, which equals $200. Academics will get some free usage, and early subscribers can lock in the current price before it rises.