Together with AI communities in Singapore, Hong Kong and the Philippines, MLT had the great honor to participate in an AI Ethics meetup initiated by Prof. Andrew Ng and his team at deeplearning.ai.
Three teams of AI Engineers and Research Scientists based in Tokyo tackled AI Ethics from three different perspectives:
- Algorithms & Tools
Our goal was to come up with an Ethics Code for AI Engineers, and vote for three actionable ways to facilitate ML Fairness: Preventing, monitoring and mitigating unwanted bias and discrimination. Ensuring that AI systems are inclusive.
- An AI engineer shall, when creating data, seek review from a diverse perspective and people impacted by their data, to avoid unintended consequences or bias.
- As an AI engineer, we need to build robust and explainable models and apply tools to visualize, making the models human-friendly.
- Also, we can associate data with models trained on it and keep track of data lineage.
The first principle was created by the group as an analogy to code review. What applies to general software engineering practices should be applicable for the dataset design, annotation schemes, etc.
The second principle refers to finding concrete ways of explaining and interpreting the predictions of black-box models, this could be in the form of visualization, surrogates model, etc.
The third principle puts emphasis on data tracking, with pre-trained models and weight sharing, as well as changing versions of dynamic datasets it’s important to incorporate a version control practice for data, just as we would do for code bases.
A team of Research Scientists led by Prof. Rei Akaishi who is leading the Social Value Decision Making Lab at the RIKEN CBS Toyota Collaboration Center in Japan, discussed a strategic approach in terms of corporate responsibility, education and outreach.
Andrew Ng is is one of the world’s leading Deep Learning experts, AI pioneer, and adjunct professor at Stanford University. He’s the former Chief Scientist at Baidu, he co-founded Google Brain, Coursera and deeplearning.ai. With his online courses, he has successfully spearheaded many efforts to “democratize deep learning”, teaching over 2.5 million students through his online courses.
deeplearning.ai is making a world-class AI education accessible to people around the globe so that we can all benefit from an AI-powered future.
Machine Learning Tokyo is a Japan-based nonprofit organization dedicated to democratizing AI through open education, open source and open science. As a research and engineering-focused community of more than 4,000 members we held more than 70 AI meetups in Tokyo, work on projects in AI for Social Good and publish ML research papers (e.g. NeurIPS 2019). More to read up on AI Ethics:
- OECD Principles on AI
- Understanding artificial intelligence ethics and safety (The Alan Turing Institute)
- Machine Learning Interpretability (GitHub repo with resources)
- ML Fairness (Google AI)
- Biased Algorithms Are Easier to Fix Than Biased People (The New York Times)
Thank you to our creative minds at SVSING and ON-1 Tokyo (IG @svsing.ppp and @on1tokyo) for capturing these wonderful shots!