Read a new article by Gradient's Dan Steinberg and Finn Lattimore showing how machine learning can be used for evidence-based policy (capturing complex relationships in data, mitigating bias in models and using regularisation for better causal estimates)
gradientinstitute.org/machin…
We’re recruiting for our Summer Scholar program 2020-21. Gain research experience working at the frontier of ethical AI with experienced AI researchers and practitioners. Positions in Sydney and Canberra. Applications close 15/08/2020. gradientinstitute.org/docs/G…#machinelearning
We are thrilled to announce our involvement in the Monetary Authority of Singapore's Veritas work. We are developing the methodology and metrics for measuring fairness in customer marketing in the finance industry (with @IAGAust Firemark Labs and @HSBC). mas.gov.sg/news/media-releas…
Our recent blog post on representing causal models within a standard Bayesian framework has prompted an active discussion amongst top causal inference experts statmodeling.stat.columbia.e…@Finn_Lattimore
New @GradientInst paper on new fast methods for fair regression. Useful for fast estimation of fair risk scores, credit scores, personalised payments and other applications with continuous-valued decisions. arxiv.org/abs/2002.06200
Our paper on a new way of assuring fairness for continuous-valued decisions (like risk scores, credit scores, payments, etc) has been accepted into EDSC2020. Read the draft at arxiv.org/abs/2001.06089
Our new blog post (w. @CriteoAILab's David Rohde) explains our #neurips workshop paper. We show that representing causality within a std Bayesian approach interpolates between tractable and impossible queries opening up new approaches to causal inference. gradientinstitute.org/blog/6…
This reflects our current thinking about ethical AI in general and within the Australian context in particular. gradientinstitute.org/news . We submitted it to the consultation on “AI: Australia’s Ethics Framework”, developed by @Data61news and released by @IndustryGovAU. #aiethics
We had some papers accepted into the Ethics of Data Science conf in Sydney next week. "On the impossibility of formalising fairness in ML" by @Finn71454004 and "Designing ethical algorithms has ethical pitfalls" by @tiberiocaetano and others. See you there sydney.edu.au/data-science/n…