Fantasy game environments could become a little more exciting thanks to AI. Researchers at Georgia Institute of Technology and Facebook AI Research have explored the possibility of equipping AI agents with NLP capabilities so that they can speak with other characters and complete desirable actions within game environments.

Their academic paper has revealed that the two approaches can produce game characters that speak and behave in ways consistent with overall motivations.

"Agents that communicate with humans and other agents in pursuit of a goal are still quite primitive," Prithviraj Ammanabrolu, one of the researchers who carried out the study, told TechXplore. "We operate based on the hypothesis that this is because most current NLP tasks and datasets are static and thus ignore a large body of literature suggesting that interactivity and language grounding are necessary for effective language learning."

“Interactive narrative games are simulations in which an agent interacts with the world purely through natural language—-'perceiving,' 'acting upon' and 'talking to' the world using textual descriptions, commands and dialog," Ammanabrolu said. "As part of this effort, the ParlAI team at FAIR created LIGHT, a large-scale, crowdsourced fantasy text adventure game where you can act and speak as a character in these worlds. This is the platform on which we conducted our experiments.

One of the earliest ways of training AI agents is to get them to practice skills within interactive simulated environments. Interactive narrative games, (text adventures) are particularly useful for training both goal-driven and conversational agents, for they enable a range of verbal and action-related interactions.