AI, machine learning are fueling advances in chatbot technology

With the popularity of chatbots comes a flurry of competing application frameworks. Oracle believes adhering to industry standards will prevent fragmentation and incompatibility.

As chatbot technology -- software that simulates conversational spoken or textual human interaction -- becomes more commonplace, application developers have yet another area in which they need to acquire expertise. With more than a dozen frameworks now available on which to build chatbots, competition is driving a rapid expansion in capability and reach.

One company providing a technology platform for chat is the database giant Oracle. Amit Zavery, senior vice president and general manager of integration products for the Oracle Cloud Platform, recently explained the company's positioning in an exclusive interview with SearchCloudApplications.

From Oracle's perspective, what is the current state of chatbot technology, and what pieces of the puzzle are missing or not yet fully mature?

Amit Zavery: A lot of the technology is in place to do natural language-based interaction to chatbots. You can integrate into messaging platforms. What's missing is some standardization. There is still fragmentation among frameworks and that needs to be consolidated and made more clearly definable across multiple platforms.

Where does artificial intelligence come into play for Oracle's chatbot efforts?

Amit Zavery, senior vice president and general manager of integration products, Oracle Cloud PlatformAmit Zavery

Zavery: Oracle's strategy is to have capabilities for modern development that is based on new technologies, languages, cloud-native development environment, and integrating everything together including sensors, systems and mobile devices. To do all this, we've built out our platform with capabilities for AI and machine learning, and a mobile front end. We are now enhancing those interfaces and experiences using chatbots.

What is the connection from AI and chatbot technology to the applications that use them?

Zavery: Around the idea of AI and machine learning, we have been delivering what we call smart applications. We have embedded machine learning algorithms and AI systems into our database, management products, as well as applications for many years. In smart applications, we have thought through the question of whether you take all of the data you are collecting, how to predict which information might be required for a user, and determine what is missing and what is needed. From that, it's possible to infer and predict customer behavior. We have built out end-to-end smart applications, which we deliver in a data cloud where we can do personalized and targeted marketing, commerce and analytics. That's available today.

Oracle has a chatbot framework, Microsoft has one, AWS, Facebook, Google, and there are more than a dozen others. Are we doomed to the same fragmentation and incompatibility we saw years ago with various flavors of Unix?

It is up to all of the providers of chatbot technology, including Oracle, to adhere to industry standards.

Zavery: I sincerely hope not. There will always be some fragmentation. There will be specific, different implementations and frameworks; developers will have to decide which to use. I think the world has evolved and recognizes the need for interoperability and consistency. It is up to all of the providers of chatbot technology, including Oracle, to adhere to industry standards, contribute back into the community, and collaborate across different systems to make sure things work.

Does that mean Oracle is not working in a vacuum?

Zavery: When we look at our chatbot technology, we work with several messaging services providers to make sure we can integrate and interoperate with Facebook Messenger, WeChat and others.

Users don't have to be concerned about frameworks, but developers do. What advice do you have for them?

Zavery: I can see that developers would be concerned about the frameworks. Our chatbots are an API implementation. When you are integrating with it, you should not have to worry about the nitty gritty of the implementation. As long as interfaces are clean and well-defined and you know what you are trying to get out of those chatbots or mobile interfaces you should be ok. Even if you rewrite the application, you are still writing to an API set without having to rewrite everything around that.

Joel Shore is news writer for TechTarget's Business Applications and Architecture Media Group. Write to him at [email protected] or follow @JshoreTT on Twitter.

Next Steps

Are you ready for specialized chatbots?

Learn how software interacts with conversational languages

Can chatbots play a role in the HR department?

Chatbot technology to follow HubSpot's Motion AI acquisition

Why more chatbots could be a good thing

Dig Deeper on Big data, machine learning and AI