Sara Hooker, formerly VP of AI Research at Cohere, is embarking on a new venture focused on developing AI models capable of adapting to changing environments. This approach contrasts with the current trend of simply scaling up AI models. Hooker's work has focused on model efficiency training techniques and optimising for models that fulfil multiple desired criteria -- interpretable, efficient, fair and robust.
Adaptive AI models are designed to learn continually, updating their knowledge in real-time as new information becomes available. This allows them to stay relevant and respond effectively to unforeseen situations. Such models are particularly useful in dynamic contexts like robotics, autonomous navigation, climate modelling, and financial trading, where environments are constantly evolving. Agentic AI systems can use generated content to complete complex tasks autonomously by calling external tools.
Hooker's new startup will likely focus on creating AI systems that are not only more versatile but also require less data to train for new tasks. This approach could have a significant impact on the development of AI solutions for various industries, especially those dealing with rapidly changing conditions.
