AI Predicts Startup Success and Simulates Platform Economies
Researchers use reinforcement learning to reveal how revenue-maximizing platforms impact social welfare during economic shocks, while another AI system predicts which startups will receive funding.
Researchers utilize reinforcement learning to simulate two-sided marketplaces like Uber and DoorDash, revealing how these platforms impact the broader economy. By testing 15 different market settings, the AI system shows that platforms designed to maximize revenue tend to raise fees and extract more profits from buyers and sellers during economic shocks, which ultimately harms social welfare.
The study builds on Salesforce’s open source AI Economist research environment to understand how platform design objectives shape market dynamics. Interestingly, the simulations indicate that when platform fees are fixed through regulation, a platform's revenue-maximizing incentives generally align more closely with overall economic welfare.
In a separate demonstration of AI versatility, researchers from ETH Zurich develop a system that accurately reads tree heights from satellite images. Meanwhile, another group of researchers tests a predictive AI model that analyzes public web data to forecast which startups will successfully receive funding, showcasing the broad practical applications of machine learning across both environmental and business sectors.