Why Better Conversations Need and How AI Can Help
Communication is the foundation of society. Communication is how we build trust, lasting relationships, professionally and politically. The conversation is more than just a perk for good company. It’s the lifeblood of every interaction.
The majority of technological breakthroughs in the past two hundred years have been focused on keeping people connected and providing a more personalized set of communication tools. We have been so focused on connecting people as fast as possible that we rarely stop to evaluate the quality of conversations.
Why better Conversations Matter and How AI Can Help
What is the secret to a great conversation? How can you quantify something so vital, but for millennia it has been a qualitative observation, almost completely subjective to those involved in the conversation?
Let’s examine the importance of good conversation and how to measure it. Then, let’s look at what we can do to improve it.
How to Quality Conversational Value
Quality conversation refers to the balance between two or more people (Cornell University). It is about balancing between too much detail or over-simplification, the flow of information within a subject area, who is asking the questions, and who answers them.
The quality of the conversation may depend on its context.
Quality can be measured by the quality of your conversation with a friend. A quality conversation in business is very different. It often focuses almost exclusively on the outcome of the call.
Is there a way to measure the objective quality and report on it?
Five Key Elements of effective conversation.
Harvey Deutschendorf recently outlined five key elements for effective negotiation at Fast Company:
- Make it about the other person.
- Practice active listening.
- Move the conversation to a deeper level.
- Ask good questions.
- Consider time and space.
Each of these touches on the attention that is paid to another party during a conversation, as you can see.
You can adjust your contribution by asking yourself these questions while you are in the conversation.
Did you hear them? Did you feel heard? Are you feeling heard? Are both sides asking strong questions? Did you adjust the pace and depth of the conversation to fit the time and space available?
Why are conversations difficult nowadays?
These questions are difficult to answer in 21st-century society due to the fragmented conversations. At any given time, you have access to 80% of the people you know.
It is possible to call, text, send an instant message, or email anyone anywhere in the world at any given time. Conversations are often shortened. Many times, conversations are not the same as they were 20 years ago.
When was the last time you called someone to ask an easy question?
Many people are becoming rusty as a result. Conversations can be more difficult to hold and this affects both personal relationships as well as business interactions. Even professional customer service agents, who are on call for 40 hours per week, have trouble getting the parts of speech and conversations right.
AI is addressing the discrepancies and lag-time in conversations.
Smart replies have been a significant part of the response to the pandemic, especially during the crisis. A recent study in Computers in Human Behavior found that participants trust AI systems more than the people they are communicating with.
We assume that conversational problems were a matter of individual responsibility.
However, the problem with these systems is that the majority of the design work involved in them focuses on the user interface and not the impact on the conversation.
This is true for virtually all the tools that we have available to help us address technology-related conversations.
AI Challenges in Good Conversational Improvements
Researchers presented specific problems in natural language generation tasks in a paper they presented at the 2019 NAACL conference. Specifically, they outlined how less open-ended tasks like machine translation and sentence compression that offered mostly word-level decisions, and for which control is less important, were more excited than open-ended conversational elements.
However, these open-ended elements are where most human conversations thrive and develop. These include story generation, abstractive summarization, and chitchat dialog.
Why are open-ended more complex than close-ended elements?
These machines require higher-level decisions, and there is usually no “correct answer.” This means that machine control is essential and that it can ask and answer questions such as “What should we talk next?” In a way that encourages quality conversation.
The researchers identified six aspects of conversation that can be used to judge chatbot quality in order to evaluate it for their experiment.
- Is the bot able to avoid repeating himself?
- Do you find the conversation to be interesting?
- Is the bot able to understand the conversation?
- What level of fluency does the bot have in the common language of conversation?
- Is the bot able to listen and respond appropriately?
- The bot is so curious!
These elements are closely related to the five components of a good conversation, as outlined in Fast Company’s article. How can an AI system seem engaged and interested in the content being discussed?
Measure and respond to human emotions
These challenges can be addressed by ensuring that AI systems are able to detect and respond to human emotion. Are humans interested in AI systems? Are you interested? Are you eager to learn? Are you becoming frustrated? These elements can all be measured with emotional AI in the human voice, which can make bots and voice assistants more engaging.
How to Improve Conversation
How can we improve communication if technology is only designed for the user?
AI & ML tools can train customer service
There are AI and ML tools that can do this. Artificial intelligence suggestions are already a part and parcel of everyday life. We are receiving recommendations almost daily, from Google’s search results to Siri’s and Alex’s automatic suggestions on our smartphones, about where to eat, what to listen to or watch, and more.
The same concept can be extremely valuable in conversation to help and improve our communication. AI-mediated conversations can be used in healthcare to augment the way healthcare professionals communicate with people with autism and mental health problems.
AI is used in palliative care to determine the key elements between patient and practitioner and then create the perfect match.
How do you respond to specific questions?
The quality of conversations can greatly improve by measuring how people respond to different types of questions, their conversational style, cadence, tone of voice, and conversational style.
This is a great tool for healthcare and customer service centers, where the right match between customer agents can improve outcomes while expediting service.
It can also be used in dating apps, where the whole business model relies on finding compatible matches between people. Even casual conversational tools such as instant messengers.
AI-Mediated Conversations (AIMC) are used to automate call routing to agents using emotion AI and voice data. Based on the observed data, they match customers with the right customer service agent.
A conversation that matches the customer with the customer service representative will result in better outcomes. You can reduce costs by improving your handling times and first-call resolution rates.
Mental health and performance
Another way to support agents’ mental health and performance is to reduce the number of difficult calls they have.
Measuring and Responding to Challenges in Conversations
To improve business and personal relationships as well as healthcare, we are able to quantify the quality of good conversations.
Miscommunications and incorrect matches account for billions in lost productivity every year. The ability to assess and respond to personality data using behavioral analytics and voice data is a major leap forward in improving the quality of conversations.