Conference: Workshop on Experimental Designs in the Age of Artificial Intelligence

The workshop "Advancements in Experimental Design in the Era of AI" will be held from March 7-9, 2025, at the UC Berkeley campus Alumni House. This workshop aims to unite experts from various disciplines to develop and discuss recent advancements in experimental designs that combine classical approaches with cutting-edge artificial intelligence (AI)-driven techniques.

In today's data-driven world, understanding cause-and-effect relationships is essential across fields such as healthcare, social policy, and industry, and randomized experiments serve as the gold standard for establishing causal relationships by systematically testing the effectiveness of treatments, interventions, or policies. Well-designed experiments not only yield reliable insights but also reduce costs, accelerate outcomes, and enhance public benefits.

However, designing experiments that meet the complex needs of different fields, each with unique challenges and data requirements, is a significant task. While traditional experimental designs have been successful for decades, recent advancements in data collection and AI potentially offer new opportunities to enhance experimental efficiency and insights.

This workshop aims to foster collaboration among researchers across different fields to create new experimental design strategies suitable for today's complex data environments.

The workshop aims to address the need for a unified approach to experimental design by bridging classical design of experiments (DoE) and modern adaptive methodologies, including reinforcement learning and AI-assisted designs. Classical DoE has been foundational in manufacturing, engineering, and quality control, emphasizing optimized balance and limited sample sizes.

However, recent applications in clinical trials and digital platforms may require more adaptive approaches that dynamically adjust based on accruing data. These modern adaptive strategies — such as response-adaptive randomization, enrichment designs, micro-randomization, and multi-arm bandits — offer enhanced statistical efficiency and personalization but necessitate tailored statistical frameworks and causal inference methods.

Despite their potential, the application of modern designs has been hindered by limited cross-disciplinary dialogue and implementation guidance. This workshop will convene experts in statistical design, biostatistics, econometrics, political science, and industry to foster interdisciplinary innovation in experimental methodologies. Objectives include fostering knowledge exchange across fields, advancing the integration of adaptive and classical designs, and applying AI tools to optimize experimental processes.

By addressing practical challenges and promoting collaboration, the workshop aims to advance experimental design theory and practice, leveraging AI to tackle the complex data landscapes of modern research and industry applications. For more information, please visit the workshop website at: https://www.design-ai.site/Berkeley-2025/ .

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.