ACM Talk: The Limitations of Data, Machine Learning & Us with Dr. Ricardo Baeza-Yates

ACM BPDC hosted an online ACM talk session titled “The Limitations of Data, Machine Learning & Us” on Tuesday, September 15th, from 7:30 PM to 8:30 PM. The talk was delivered by Prof. Ricardo Baeza-Yates, a WASP Professor at KTH Royal Institute of Technology in Stockholm, and a part-time professor at both Universitat Pompeu Fabra and the University of Chile. As a recognized ACM and IEEE Fellow, former VP of Research at Yahoo Labs, and an expert in responsible AI, Prof. Baeza-Yates shared his extensive expertise on whether machine learning is being used ethically and scientifically. The virtual event was open to all, and free certificates were provided to all participants. During his presentation, Prof. Baeza-Yates challenged the ubiquitous applications of machine learning, conducting a deep dive into the specific limitations of supervised ML and its core component, data. He explained the inherent limitations of data by exploring issues such as small data, datafication, bias, predictive optimization, pseudoscience, and the problem of evaluating success instead of harm. To illustrate the broader context of these issues, he also examined our own human limitations when utilizing machine learning. This included an examination of various types of human incompetence, ranging from cognitive biases, unethical applications, and administrative incompetence to misinformation and the impact of ML on mental health. Also, to help researchers and designers mitigate the problems outlined above, Prof. Baeza-Yates introduced a comprehensive discussion on AI regulation and responsible AI principles. Ultimately, the keynote offered a compelling critical perspective, demonstrating that the tech community must actively acknowledge the limitations of both data and human behavior while applying responsible AI principles to navigate the modern machine learning landscape.