Ganes Kesari的动态

Should you care about AI #explainability? On the one hand, enterprises continue to favor simpler ML models that are easier to implement and explain. But then, we have black-box AI models silently getting rolled out en-masse. There are many examples. This very post showed up on your Linkedin feed over hundreds of posts, from your other connections. We can’t tell *exactly* why. So, is the case with your ‘personalized’ feeds on Facebook, Twitter, news apps or any tool that promises #personalization. Every #recommendation algorithm that ‘works well’ is most certainly a #blackbox. We seem to have shifted our focus to ‘game’ these systems and ‘amplify’ our reach. Are we getting inured to non-explainable AI? Are we already too far? Is explainability another early-stage ‘adoption’ hiccup that is slowly fading away? What do you think? What are some popular AND transparent algorithms that come to mind? #datascience #artificialintelligence #transparency Pic source: https://xkcd.com/1838/