Larry Lewis
AI STRATEGY
Larry Lewis
Larry Lewis spent a decade analysing real-world operations as the project lead and primary author for many of DOD ' s Joint Lessons Learned studies . For example , he was the lead analyst and co-author ( with Dr Sarah Sewall ) for the Joint Civilian Casualty Study ( JCCS ). General Petraeus described the study as " the first comprehensive assessment of the problem of civilian protection ." issue can originate from one of two ways . An AI algorithm can contain data bias due to the algorithms being trained using biased data . The algorithms can also contain societal AI bias where assumptions and norms within society can cause analysts to have blind spots when it comes to analysing datasets .
Discussing AI bias and policy , Lewis said : “ We need to be mindful . We need to be deliberate about both understanding potential biases and what those effects will be . It ' s also useful to remember not all biases are bad . You can actually build in biases for certain applications to actually get better performance . However , we need to be deliberate about addressing biases that will lead to bad things .”
Lewis outlined the steps that needed to be taken to ensure issues around regulation , ethics , privacy and equality are balanced whilst promoting innovation in this space . She said : “ Policymakers should not view regulation as presenting an “ either / or ” choice between promoting innovation on the one hand and fostering ethical use , fairness and equality , and privacy on the other . Instead , they should strive to craft regulations that accomplish both goals . For example , organisations are more likely to purchase and adopt AI products if they know that they can trust them . Individuals are more apt to use AI products that they can trust .”
Concluding , Lee stressed the importance of aligning both innovation and regulation as it will , in turn , create a better technology system for government organisations : “ Recognising that these goals are complementary and not mutually exclusive provides a good framework for crafting risk-based and proportionate regulations and standards that promote fairness , explainability , transparency , safety , and accountability as well as innovation .”
aimagazine . com 53