News dalla rete ITA

24 Novembre 2020



Businesses that use artificial intelligence systems to make decisions involving customers risk breaching existing anti-discrimination laws, the Australian Human Rights Commission has warned, in a paper trying to help lift the quality of algorithms and help Australian developers compete globally.With more AI being deployed across the economy, the commission and a group of research partners say some AI models risk falling foul of age, race and sex discrimination legislation. The technical paper offers suggestions for avoiding so-called algorithmic bias, such as boosting data pre-processing and model complexity.Produced by the AHRC with the Gradient Institute, Consumer Policy Research Centre, CHOICE and CSIRO's Data 61, the paper provides a clarion call to company boards to take responsibility for mitigating legal risks associated with AI. It also suggests regulators carefully monitor the outcomes of automated decision-making, and says fintech companies must be aware that they operate within existing legal frameworks protecting fairness.Banks, insurance and energy companies are called out for failings in the report, which uses simulations to show how training machines to focus on profitable customers can introduce problematic bias.The new technical paper comes after the commission's December discussion paper found ethics frameworks developed by global tech players were not enough to address the problems, and that the law itself needed to set the baseline for proper conduct. A final paper on its AI project is due to be released early in 2021. Despite the warnings, the new paper also strikes an optimistic note, suggesting ethical design of algorithms could help Australia compete globally by creating fair and respected AI products for export. To build ethical AI systems, any moral considerations must be explicitly represented in the objectives, data and constraints that governed how AI systems made decisions.Commonwealth Bank has appointed AI expert Professor Genevieve Bell as a non-executive director, but few other large companies have any AI expertise in the boardroom. After the Hayne royal commission into the banks, the Human Rights Commission is also pushing AI users to go beyond the law when reducing unconscious bias risk.The high level of focus on ethical and legal AI application comes as business leaders including mining magnate Andrew Forrest, through his Minderoo Foundation, are seeking to influence AI governance.Global tech giants Microsoft and IBM are also battling to win over customers by meeting demanding community expectations on privacy and the explainability of decisions. Meanwhile, a World Economic Forum study of AI application in financial services last year predicted it will fundamentally change the role of prudential supervisors as banks reinvent themselves as hubs for system-wide intelligence. (ICE SYDNEY)

Fonte notizia: Financial Review 24.11.2020