NeurIPS 2020 Highlights Push for Responsible AI and Real-World Impact
This year's virtual NeurIPS conference emphasizes that AI researchers must take responsibility for the societal impact of their algorithms. Speakers urge collaboration with other fields to ensure technology does no harm and supports vulnerable data workers.
The annual NeurIPS conference attracts more than 20,000 virtual participants who focus on the growing need for responsible artificial intelligence. Keynote speaker Charles Isbell delivers a powerful message that machine learning researchers must stop ignoring the real-world consequences of their work and stop passing the blame to other scientific disciplines.
Isbell urges developers to adopt a systems approach that questions how technology operates in society, who uses it, and what could go wrong. To answer these critical questions, AI scientists need to collaborate extensively with other stakeholders rather than working in isolated academic silos.
Other speakers echo this theme by highlighting the plight of low-paid data labelers and exploring ways AI can help these invisible workers unionize or build new skills. Additionally, researchers discuss the importance of causal relationships in data to ensure artificial intelligence functions safely and effectively in practical applications.