We have already seen how clumsy computers can be when they try to speak like humans, but researchers at North Carolina State and Georgia Tech have developed a program that could make it easier to show them how it’s done. Their approach, described in a recent paper, allow developers to build natural language generation (NLG) systems two times faster than currently possible. NLG technology is used in a wide range of applications (including gaming centers and customer service), but production of these systems has traditionally required the developers to enter massive amounts of data, vocabulary and models – the rules that computers use to develop coherent sentences. The lead author, Karthik Narayan and his team, however, have created a program capable of learning to use these models on its own, which requires developers to input only the basic information on a topic of conversation. As he learns to speak, the software can also make suggestions on automatic information should be added to its database, based on the conversation at hand. Narayan and colleagues present their study artificial intelligence this year, and Interactive Digital Entertainment conference in October, but you can dig through it for yourself, from the link below.