Facebook has built a new AI framework tool that's designed to help build and enhance conversational AIs, called ParlAI. The tool essentially allows developing conversational AIs to train on a large number of different models and protocols simultaneously, significantly accelerating growth and helping the AIs expand to be able to learn more things faster. It comes with a wide range of pre-made example models to train on, built on top of frameworks like Google's TensorFlow, and the LUA programming language. The open-source framework is available on GitHub for anybody that may want to fork it and try their hand at building a more customized version.
While the endgame for the project is to help make any AI that has to converse with humans far better at that part of its job, some of the sub-goals are fairly lofty, too. One of the biggest is to provide a de facto, unified framework and gold standard to judge conversational AIs against. The creation of such a standard, obviously, requires everybody in the field, or at least close to everybody, to ditch their own measurement tools and standards in favor of ParlAI. This isn't as far-fetched as it may sound, though; ParlAI is not only open-source and easily obtainable, but links seamlessly with Amazon's Mechanical Turk service, which means that programmers can have humans interact with and evaluate their AI for a small fee very easily. It also runs Amazon Web Services as a backend, making it incredibly easy to implement in almost any stack. The whole thing is built on Python, one of the easiest and most common programming languages out there, found in web and native apps as high up on the totem pole as Google.
Installing ParlAI to start playing with it and prepping it to implement into your own stack happens in only two steps; simply clone the GitHub repository, then enter the cloned directory and start messing around. Being open-source, more niche variants of ParlAI are bound to pop up with time, which will help it to grow a community organically. ParlAI has a real shot at achieving its goal of becoming the go-to standard for conversational AI, but it has to deal with competition from a number of larger incumbents to do so.