AI & ML interests
ML Safety
SafeML
About
We Make AI Safe
At SafeML, we envision a future where artificial intelligence (AI) is as ubiquitous and essential as electricity is today. Our company was founded on the premise that, in the 21st century, the seamless integration of AI into every aspect of our livesāfrom our cars to our hospitalsāmust be characterized by unwavering safety and security. We aim to enhance human well-being by preventing AI-related risks such as traffic incidents in self-driving cars and diagnostic errors in medical imaging, ensuring that AI safeguards rather than compromises our quality of life.
Focus Area
Robustness within computer vision
Hendrycks et al., 2021 identified four critical areas ready for ML Safety research: Robustness (withstanding hazards), Monitoring (identifying hazards), Alignment (steering ML systems), and Systemic Safety (reducing deployment hazards).
Building upon our founders' experience in automated driving and medical image analysis, we strategically concentrate on 'Robustness' within computer vision, with a particular focus on resilience against long-tail events, often referred to as 'Black Swan' events. Our strategy involves creating tangible products and tools specifically designed to enhance this aspect of robustness, enabling ML developers to increase the safety of models that are deployed in the real world.