Although Hollywood’s portrayal of artificial intelligence (AI) in movies such as War Games, Terminator, 2001: A Space Odyssey, Her and A.I. is entertaining, it doesn’t accurately mirror AI or chatbots in real life. So began Microsoft Cloud Solutions Architect Wayne Smith, as he walked through the real nuts and bolts of bots and AI during the session, Serverless Computing and Chatbots with Microsoft Azure.
Responsible artificial intelligence
One of the main differences between Hollywood AI and real-life AI, Smith pointed out, is that the movies don’t portray responsible AI, which he encouraged and reinforced during the session. What, then, are the rules of responsible AI? Here are a few:
- AI must be designed to assist humanity
- AI must be transparent; a bot should always admit it’s a bot
- AI must maximize efficiencies without destroying the dignity of people
- AI must be designed for intelligent privacy
- AI needs algorithmic accountability, so humans can undo unintended harm
- AI must guard against bias
With these responsibilities in mind, organizations can create a web app bot using Microsoft Azure’s QnA Maker. But before building a bot, several options and capabilities should be considered.
Considerations in a bot build
One consideration in building a bot is the means. Users interact with bots through different means, such as text messages and voice. Voice transfers information faster, making it the preferred option.
Another consideration is the range of capabilities you want in your bot with four different types: knowledge, application enhancements, functional and virtual assistants.
Knowledge bots can replace the FAQ section of a website. They can streamline the Q&A process and answer general questions.
An application enhancement can be added to an already existing app that doesn’t already feature conversational input. The enhancement provides contextual information and assists users in everyday tasks.
With a functional bot, the bot is the app. It can talk to other systems and expose siloed information.
Virtual assistants, such as Cortana, Alexa, Google and Siri, are most familiar to consumers. They perform a variety of functions and provide true contextual knowledge, such as being able to answer questions by knowing the user’s location or calendar.
The end user + bot interaction
As you build your bots, you must also consider the conversational concepts, also known as “turns,” which are the interactions between the bot and the end user. In a single turn, the bot answers a question. In a multi-turn, the bot asks follow-up questions before giving an answer, resulting in a conversation.
You can create natural language processing in your bot using Azure, which offers an open and extensive platform that allows you to build a simple or more complex bot based on your needs.
AI and chatbots are becoming increasingly ubiquitous as more businesses incorporate them into customer service options and consumers rely on the conveniences they provide. As this popularity continues to rise, we’ll see more and more AI and chatbots injected into our daily lives.
Learn more about Microsoft Azure and building bots today. Plus, stay tuned for the next AI developments coming from Microsoft at Cloud Summit 2020 from May 11-3 in Miami, FL.