[ad_1]
In July, Uber released a new open source artificial intelligence library called the Plato Search Dialogue System. A few months ago, Cisco opened its MindMeld Conversational Artificial Intelligence Platform, after acquiring the company of the same name in 2017 for $ 125 million.
Why are so many new libraries being announced? There seems to be a tendency for developers to embrace open source conversational AI and leave behind closed alternatives. Most of the key tools developed over the past three years to develop conversational artificial intelligence come from open sources. And companies like Uber and Cisco want to be able to set the standard conversational AI stack.
In fact, the entire field of AI has seen a strong shift towards open source infrastructures in recent years. The initial spark may have been Google's decision to open the TensorFlow open source software in 2015. At that time, many companies began to pay attention. Data from Google Search shows that interest in open source libraries such as TensorFlow and PyTorch is increasing at the expense of closed platforms such as IBM Watson and Sagemaker from Amazon.
The market is driving this change. Companies are increasingly deciding that many of the AI capabilities they need are strategically important and need to be developed in-house. Using open source tools, they can create their own sets of training data and other IP addresses, such as custom integrations with their back-end systems. By developing the talent, data, and software to ship the AI themselves, these companies control their own AI destiny.
This trend has now reached the field of conversational AI. By examining developers' interest in two of the most popular cloud tools, it is clear that the number of questions asked about Stack Overflow has not changed since early 2018. (Note that although botframework is the name of 39; an open source SDK, it is used to access Microsoft's conversational AI platform, which is not open source.)
While the interest in closed source tools, such as Dialogflow and Microsoft Bot Framework, is declining, the monthly installs of the most popular open source natural language comprehension (NLU) libraries have been multiplied by six between mid-2018 and mid-2019. I predict that within 12 months, the open source will surpass the cloud APIs and will become the dominant force of conversational IA.
So why now? In 2016, Chatbot's hype reached its peak, with companies exploring chatbots and voice assistants. To establish a proof of concept, the solution of a fully hosted solution such as Dialogflow is compelling because it requires very little engineering effort or upfront costs. Today, however, companies in different industries are deploying conversational artificial intelligence to solve more complex problems, and many prefer to control tools and train themselves.
The conversational AI market is beginning to mature, particularly in the banking, insurance and healthcare sectors. Erica, of Bank of America, and Eno, of Capital One, are examples of large banks that have formed large teams to develop conversational AI. Challenger startups such as Lemonade and N26 are about to create stand-alone organizations, which becomes possible as the industry moves from level three conversational conversational intelligence to level five.
As always, with the maturity of the market come differentiated products. There are already many competing open source solutions for different types of developers. The main goal of MindMeld is to enable developers to create use cases on Cisco-specific conferencing devices and the broader Cisco ecosystem. Uber's Plato system operates in another vertical; it is optimized for the needs of researchers who want to develop and test new algorithms and evaluate them on standard data sets, similar to other research-oriented libraries, such as PyDial and the very recent ConvLab. Then, of course, tools are intended for developers who send products in production. Suppliers include Rasa (where I am CTO and co-founder), DeepPavlov and Snips.
If the open source exceeds the cloud APIs over the next 12 months to become the dominant force of conversational AI, Cisco and Uber tools are only a beginning.
Alan Nichol is founder and CTO of Rasa.
[ad_2]
Source link