Conversational UX - Politechnika Gdańskamanus/SAB/Wyklad_chat.pdfConversational UX: Agenda...

Preview:

Citation preview

Conversational UXDialogflow in action

Conversational UX: Agenda

● Background, history and future ● Most popular frameworks● Key components of C-UX system● Introduction to DialogFlow● Demo● Lab introduction

Chatbots - the beginning

Chatbots - text games

Chatbots - more recent progress

Chatbots - more recent progress

Chatbots - more recent progress

Chatbots - more recent progress

Chatbots - failure of the Tay experiment

● Tay was AI chat bot released by Microsoft via Twitter● It was supposed to mimic casual, jockey speech patterns● The bot began to post inflammatory and offensive tweets● Microsoft was forced to shut it down only 16 hours after its launch

See: Tay's historyl

Dialog based UX is often used in video games

Chatbots - predictions by Gartner

Consumers Favor Visual and Voice Search

By 2021, early adopter brands that redesign their websites to support visual and voice search will increase digital commerce revenue by 30%.

With visual and voice search rapidly increasing in popularity and on the way to being dominant mobile search modes, enterprises should experiment to identify the best ways to capitalize on this consumer shift. This search type enables marketers to gather more robust information about consumer habits, and early responders will see an increase in conversion rates, revenue, new customers and customer satisfaction.

Chatbots -predictions by Gartner cont...

Bots Take Over

By 2021, more than 50% of enterprises will spend more per annum on bots and chatbot creation than traditional mobile app development.

Individual apps are out. Bots are in. In the “post-app era,” chatbots will become the face of AI and bots will transform the way apps are built. Traditional apps, which are downloaded from a store to a mobile device, will become just one of many options for customers.

Chatbots -predictions by Gartner cont...

AI Creates More Jobs Than It Takes

In 2020, AI will become a positive net job motivator, creating 2.3M jobs while only eliminating 1.8M jobs.

AI will have a positive effect on jobs with AI-related jobs seeing steady growth starting in 2020. Up through 2020, all industries will see varying levels of time and effort savings with most industries experiencing no job loss. More than replacing humans entirely, AI will augment existing jobs and improve productivity.

4 major players

Amazon success with Alexa

How is DialogFlow can be used?

● Support existing application with language processing● Enhance search● Create a chatbot

○ Really just a decision tree○ Machine Learning to match inputs to given examples and extract entities○ Almost meets definition of an Intelligent Agent

■ appropriate actions for goals and circumstances■ is flexible■ can learn

Main role of Dialogflow-like systems

{ "query": "Book me a flight to Cairo", "intents": [ { "intent": "BookFlight", "score": 0.9887482 }, ], "entities": [ { "entity": "cairo", "type": "Location", "startIndex": 20, "endIndex": 24, "score": 0.956781447 } ]}

Book me a flight to Cairo

Other features and cost

● Bot/dialog creation (Dialogflow)● Extra diagnostics (Luis)

Usually those tools are free of charge for basic usage

Architectures

Azure architecture example

Main Components of Dialogflow

1. Intents2. EntitiesSupported by: Contexts, Parameters, Integrations, Fullfilments etc.

Role of Intents

● Main component of CUX application/chatbot● Represents a mapping between what a user says and what action should

be taken by your software.● Two approaches

○ Model your intents after what your user hopes to archive (i.e. setAlarmClock)○ Model your intents after what your agent hopes to receive (i.e. paymentMethod)

Intents and DialogFlow

● Training Phrases

Intents and Dialogflow

● Training Phrases● Action &

Parameters

Intents and Dialogflow

● Training Phrases● Action &

Parameters● Responses

Intents and Dialogflow

● Training Phrases● Action &

Parameters● Responses● Contexts

Entities

● Define data types for you application● Flexible naming through synonyms● Lets you extract parameter values from your natural language inputs● Build-in entities can be powerful for non-trivial extraction tasks like dates,

date ranges etc.

Other DialogFlow ingridients

● Integrations● Fulfilments

Lab

● Come up with your own idea

● Create a bot with a few intents and a few entities

● Bonus: use webhooks (fullfilment)

● Bonus: use integration

DialogFlow Demo

● Let’s create a bot for Star Wars t-shirt store● We would like to:

○ Let user pick a t-shirt with customized size and color○ Then allow user to apply Star-Wars related print on it

● We will additionally do:○ Slack integration

Recommended