Google’s Smart Compose suggests sentence completion just like how we get word completion suggestions. This feature uses language generation models and lots of data. When I was reading Google’s blog post about Smart Compose, I came across few research papers and blogs that are about data and the problems which are created when it is biased. I have summarised all of my readings in a post which got published in Towards Data Science.
Those who find time to read this blog, do let me know your thoughts and any kind of feedback is welcome. Thanks!