2016: The Year When Chatbots Were Hot
Chatbots have been around for a while. Perhaps the first chatbot of notoriety was Joseph Weizenbaum’s Eliza, introduced in 1963, a chatbot that simulated communicating with a human. Since then, chatbots have evolved and tools have become increasingly more available and powerful.
This year marked a dramatic shift in how we build and think about chatbots. Multiple vendors have built chatbot platforms, API’s are readily available, and there is a plethora of open-source AI software within reach of developers.
Rise of the Platforms
The new platform for chatbots is messaging. While developers have used social media API’s as chatbot hosts for a while, companies like Microsoft, Facebook, and Slack have begun supporting chatbots on Skype, Messenger, and Slack, respectively. This is a short list of well-known company names and looking around will show you how many of the other messaging platforms are either planning or currently offering a chatbot plaform. Many of these platforms are still in beta and evolving, based on customer and developer feedback.
In addition to messaging, voice is an emerging chatbot platform. Amazon’s Alexa took off like a rocket this year as a home appliance that you can talk to and obtain information or control other home appliances. Google Home is another voice device in this category. Microsoft recently partnered with Harman Kardon to bring the Cortana voice assistant into similar devices. Voice is quickly emerging as a useful chatbot platform.
There’s an API for That
Up until this year, developers have either used tools like AIML or built their own logic to support chatbots. By the time 2016 rolled around, many of the corporate developers building current messaging and voice platforms were solving problems with their own libraries. Some of these libraries formed the basis for chatbot platforms that emerged earlier this year. Essentially these platforms consist of servers that perform routing to manage user/chatbot communication, libraries for intercepting and responding to users, and deployment standards via webhooks or similar interfaces. With current vendor toolsets, it’s easy for developers to quickly produce simple chatbots.
In addition to platform vendors, there’s a plethora of 3rd party API’s for managing chatbots. For chatbots, a Natural Language Processing (NLP) API is a must-have tool for translating human sentences into chatbot readable instructions. Three early NLP services were LUIS.ai, Wit.ai, and API.ai. As the chatbot ecosystem continues to expand, we’ll see a growing list of API services that make chatbots smarter and their development easier.
Ubiquitous Open-Source
Open source is the win-win of modern software development. Vendors and independent developers have a continuous stream of feedback that improves quality and available feature set. Developers can use free software that eases the on-ramp to technologies that would have either been too expensive of cost-prohibitive to develop on their own. The effect is especially beneficial in the realm of Artificial Intelligence (AI) and Machine Learning (ML).
A few perfect examples of this are Microsoft’s Cognitive Toolkit, Google’s Tensorflow, and Open AI. Each of these projects are open-source AI/ML software that developers can contribute to, study, and use in their own software. In particular, the Open AI project is designed for producing open-source AI software with a focus on ethical and safe operation. The tie-in here is that chatbots are often associated with intelligence and tools like this make it easy for developers to integrate AI into chatbot logic.
Toward 2017
In 2017 we’ll see vendors offering RTM/RTW versions of their chatbot platforms or releasing new versions with new capabilities. You’ll see more 3rd party vendors offering APIs for services we haven’t thought of yet. In addition to messaging, Voice will become a major player in the chatbot space. This was such an exciting year for chatbots and I can’t wait to see what will happen in 2017.
