We are entering a new digital era – we will use less stand-alone apps and instead access apps within app platforms, like Facebook Messenger, Slack and WeChat. We will interact with apps using chat, or the Conversational User Interface (CUI). Most of these apps will be smart, nimble applications called Chatbots. Chatbots can initiate actions and respond to requests. In my new report, Chatbots: The Rise of the Conversational User Interface http://bit.ly/1MXLiOI , the impact of Chatbots is explored. Chatbots will not truly come to power until they become smart, and they become smart through the use of Artificial Intelligence (AI).
The AI game is shifting, and AI applications are poised for mass market adoption. Here is an excerpt from the Chatbot report concerning AI:
The field of AI has accelerated in the last few years. AI programs are smarter, learning faster, and becoming more affordable and accessible to developers. The most sophisticated AI is best suited for image recognition, text analysis, time-series data (like stock price action) and fraud detection. Chatbot developers are beginning to leverage these capabilities.
First, a quick primer on terms in AI:
Umbrella term used to describe the capability of a machine to imitate intelligent human behavior. In complexity, AI moves from machine learning to deep learning/neural networks.
Explores the study and construction of algorithms that can learn from and make predictions on data. The program must be “taught” by engineers to learn to perform tasks.
More sophisticated branch of machine learning in which the algorithms are better at learning, adapting on their own with less teaching from engineers.
Artificial Neural Networks (ANNs):
Machine learning inspired by biological neural networks (particularly brain), used to estimate or approximate functions that depend on a large number of inputs and are generally unknown. The most sophisticated deep learning taps into ANNs.
The most sophisticated initiatives involving algorithms that learn on their own like a human brain (deep learning using ANNs) were until recently, the purview of players like Google, Baidu, Facebook and Microsoft who have the computing power/GPUs to run them. But in the last year, there has emerged a significant movement to democratize deep learning neural networks, which means a great deal of third parties outside of the computing giants can begin to access the computing power to build smarter apps, like chatbots.
In June 2014, a startup called Skymind http://www.skymind.io launched the first commercial-grade open source distributed deep learning library written for Java called Deeplearning4J (DL4J) http://deeplearning4j.org . The initiative means that deep learning software can be built in Java and can run on top of Hadoop. Oh, and by the way, DL4J is partially funded by WeChat parent company Tencent.
Within weeks of each other in November-December, 2015 three deep neural network PAAS services were launched – Minds.ai http://minds.ai , Metamind https://www.metamind.io and Google’s TensorFlow https://www.tensorflow.org
Mark Beccue Consulting predicts smart, AI-powered chatbots will be in widespread enterprise and consumer use by mid-2017.