Artificial Intelligence Performance

Research on artificial intelligence (AI) performance currently focuses on understanding and improving human-AI collaboration, particularly addressing issues of trust and appropriate reliance on AI systems. This involves investigating human biases in evaluating AI outputs, exploring methods to calibrate user confidence in AI predictions (including the surprising finding that overconfidence can sometimes improve human performance), and developing frameworks for effective human-AI interaction across various tasks and domains. These efforts are crucial for responsible AI deployment, impacting fields like healthcare, cybersecurity, and engineering, where AI systems are increasingly integrated into human workflows.