By the end of high school, writing evaluations should be hunting for subtleties that no standardized test – by definition – will ever be able to uncover.

Today, dozens of companies offer writing-critique services online, with programs that offer to catch your errors, tweak your use of passive voice, and improve your readability.

But Google's DeepMind AI group has now developed a reinforcement learning tool that can develop extremely optimized algorithms without first being trained on human code examples.

Rick Breen in Waltham, Mass., tells us about testing the software he helped write for an electronic instrument. The engineers write code for a long time before they can actually hear what it is ...