This document provides an overview of how to make chatbots more human-like. It discusses the history of chatbots and how they have evolved from using decision trees to natural language processing (NLP). Key aspects that can help make chatbots appear more human include giving them a voice, personality, ability to engage in small talk, and emotional intelligence. The document also discusses testing chatbots on current events, small talk, questions, complexity, and empathy as well as examples of AI characters in movies. It concludes with an overview of building a rule-based chatbot with capabilities like handling conversations, requests, and integrating with Wikipedia.
2. Agenda
● History of Chatbots
● How do Chatbots work
● What makes a Chatbot more human-like
● Artificial Intelligence Characters in Movies
● How to test a Chatbot
● Demo
6. How do
Chatbots
work?
Earlier versions of chatbots followed
predetermined paths with decision-tree
answer options which the user can select
from
Natural Language Processing (NLP) is
concerned with how technology can
meaningfully interpret and act on human
language inputs.
9. What makes
a Chatbot
more human-
like
Voice (Alexa, Siri)
Characteristics (Name, Age, Origin)
Personality
Engage users with Small talk to build a
friendly rapport, eventually leading to
trust
Emotional Intelligence: Facial Emotion
Recognition using Computer Vision
Self-awareness
10. What makes
a Chatbot
more human-
like
Slow down response time
Avoid long answers
Dealing with Surprise, Happiness and
Gratitude
12. What is your favourite Artificial
Intelligence movie character?
13. How to test a
Chatbot
Current events
Small talk
What was the first question I asked?
Ability to handle complexity
Empathy Ploy (Ability to understand
emotions)
The Turing Test
The Turing test, developed by Alan Turing in 1950, is a test of a machine's ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human.
The year was 1950. Alan Turing, who during World War II was one of the most prominent breakers of German code, issued an open challenge to the computer scientists of the world. It came to be known as the Turing Test. Turing challenged scientists to create a program that would be indistinguishable from a human in a Natural Language conversation.
The Turing Test
The Turing test, developed by Alan Turing in 1950, is a test of a machine's ability to exhibit intelligent behaviour equivalent to, or indistinguishable from, that of a human.
The year was 1950. Alan Turing, who during World War II was one of the most prominent breakers of German code, issued an open challenge to the computer scientists of the world. It came to be known as the Turing Test. Turing challenged scientists to create a program that would be indistinguishable from a human in a Natural Language conversation.
Understanding the Tech
There are two basic buckets into which chatbots fall: Scripted bots and Natural Language Processing (NLP) bots. Scripted bots, the simplest form, follow predetermined paths with decision-tree answer options which the user can select from. For Financial players looking to implement more complex use cases (like customer service or advanced analytics) into their business-models however, NLP bots are the key.
Taking a look under-the-hood, NLP chatbot technology is underpinned by three basic concepts:
Intents (what the user is trying to say),
Entities (what information did the user give) and
Utterances (how users can ask for something).
The first two are used in structuring the bot, and the last facilitates training the application to improve over time. The details of the technology, while fascinating, are beyond the scope of this piece, but for a comprehensive and easy-to-read look at the magic of chatbots, Amir Shevat’s “Designing Bots” is a sensational choice.
Personality: Consider including emojis and GIFs to give your bot an extra edge.
Small talk: You don't just jump into a deep conversation. Helps to establish trust. May yield important information. Convey warmth and affection.
Emotional Intelligence: Build and emotional connection by recognising user mood and responding accordingly.
Emotion detection technology requires two techniques: computer vision, to precisely identify facial expressions, and machine learning algorithms to analyze and interpret the emotional content of those facial features.
Computer Vision: Mood Detection using Image Processing.
Computer vision models can identify important parts and features of the face, such as the eyebrows, shape of the mouth, and eyes.
Self-awareness: Transfer a conversation
Slow down: Consider including emojis and GIFs to give your bot an extra edge.
The more complex a request through a conversation becomes the less capable an Artificial Intelligent bot will be able to comprehend and respond.
Comprehension. The lack of deep understanding of intent is the reason why you do not see many conversational chatbots for service currently.
Empathy: AI is lacking in cognitive empathy because emotions between humans are really hard to understand and explain. So, intentionally creating an empathetic dialogue with your human being or AI/chatbot can be revealing.
Mitsuku, or Kuki to her close friends, is a record-breaking, five-time winner of the Loebner Prize Turing Test and the world's best conversational chatbot
https://www.pandorabots.com/mitsuku/