Menu
January 23, 2019 | By

AI IRL Podcast Episode 4: Why Natural Language Design Is the First Step to Make an Effective Chatbot


Subscribe via iTunes, Spotify and more.

In all the hype of AI and the rush to get something “out the door”, companies are starting to realize how poor AI and chatbot design can be detrimental to a business.  Having “AI” isn’t good enough.  Having an AI chatbot that can control conversations and solve problems all while keeping the user engaged and feeling great about the interaction is key.

This is where Hans van Dam’s expertise comes in.

Hans is the managing director of Robocopy and founder of The Conversational Academy. His company is changing the way AI interacts with customers by putting first thing first: natural language design. The Conversational Academy exists to train designers in this skill.

Whether you’re a UX designer, a copywriter, or a product owner, this episode is for you.

The 2 Ways Companies Get the Process Wrong.

  1. Many companies start by creating a flowchart, focusing on the business process they already have in place and knowledge management. Then they hand it to a copywriter to change the lorem ipsum into words.  In actually, you want to take it the other way around. Start with a user.
  2. Hans says instead of focusing on “what’s the next line the bot needs to say,” put two people in a room. Tell one he’s a bot and the other he’s a human. Have them play it out like improv theater. Whatever comes out is the first draft. Start with one user, what they want to achieve and how they want to feel.

Another problem many companies have is getting carried away with fancy, shiny tech.

“Before diving into the weirdest tech, just try putting together a couple conversations that work.” Hans Van Dam

One thing Hans’ company gets so right, is their 3 basic ingredients: understanding psychology, understanding tech, and writing effective copy. There are many benefits to nailing the perfect blend of these ingredients.

Natural Language Design Lets You Control the Conversation

It’s important to understand each party’s needs. Artificial and human brains have different limitations, capabilities, and triggers.

Bots need:

  • intent
  • context
  • variables
  • entities

Humans need:

  • empathy
  • guidance
  • persuasion
  • a pat on the back

Natural language creation should blend each party’s needs. Hans emphasizes creating methods for each different conversation type.

A customer service conversation has its own structure: the AI needs to do intent recognition, then try to solve a problem. In another case, maybe a financial institution is using a bot to try to persuade the user to update his payment information. This is a very different conversation in which the bot is asking tougher questions about a person’s finances.

“Being very thoughtful about conversation design allows you to take control of the conversations.” Hans

If you control the conversation you depend less on your AI. It’s not AI vs. conversation design. They need each other.

What about Buttons?

If you can solve an issue without buttons, then you know you have a perfect conversation. Hans emphasizes voice as his primary focus; however, there are some cases where buttons can be very effective.

Buttons can be used to steer the user’s thoughts. Maybe you want your user to give a positive answer; if you have 4 buttons and 3 are positive, the odds of a positive answer increase. This can work well for persuasion conversations like sales or booking meetings.

It can be good to have buttons, but it depends on the use case. In most cases, the goal is to push conversation design to try to solve without buttons.

4 Reasons Conversation Design Needs to Focus on Empathy

Every conversation involves the feelings of the human user. Emotions should not be ignored. Hans gives four reasons he focuses intently on integrating empathy into each conversation.

  1. It’s the hardest thing to get right.  Hans utilizes a bot scorecard that grades for 5 things: personality, natural language, helpfulness, persuasiveness, and empathy. Empathy is always the trickiest one.
  2. Even though users know a bot isn’t really “sorry,” they want to feel understood.  Often this can be done without simulating emotion. A bot can say “Ok, so you’re stuck on a train, and therefore you’re going to miss your flight.” Without feelings, the bot is understanding what’s going on, which makes the user feel listened to.
  3. Empathy helps with intent recognition.  If a bot says “Ok, so such-and-such thing is going on” and the user confirms, the user feels understood, and the bot knows which dialog to run.
  4. It prevents escalation.  Empathy is about addressing the softer side of the brain to keep them engaged and motivated. Keep them interacting with the chatbot.

“Empathy isn’t just saying ‘I’m sorry,’ it’s showing you understand the user and his context. Hans”

Fearful users can become inactive, if a process feels overwhelming, these are the kind of people who allow envelopes to go unopened and pile up. Angry users can become vengeful; they write negative reviews and posts on social media telling others to stay away.  Copywriting that uses empathetic psychological techniques like depersonalization, positive framing, social proof, and expectation management can prevent these frustrating situations.

Hans gives one caution here though. Companies can become over-obsessed with sentiment analysis. At the end of the day, when there’s something negative happening to a user, the person is upset and you need to deal with it. They care about empathy but they really just want the problem solved.

Conversational Academy

Robocopy started out as a small agency; they trained their own people in their own in-house academy.  People were emailing Hans all over the world asking to get into it.

So they launched a webinar series that was a huge hit.  Now they’re putting together a full course, over 10 hours of video in which people can be trained and certified conversation designers.

His program teaches:

  • natural language,
  • cooperative principles,
  • persona development,
  • scenario development,
  • conversation design,
  • sales and persuasion,
  • managing emotions,
  • optimization
  • copywriting techniques

It launches in January

Alongside the program is a subscription model for continuous improvement.  They plan to teach conversation design that’s specific to certain platforms, talk about trust and AI, have guest lectures, and much more.

“Everyone’s building hammers and we’re the only one’s training carpenters

This AI discussion with Hans van Dam was taken from our podcast. If you want more AI: In Real Life, check out our podcast on iTunes.


Discuss / Read Comments

Leave a Reply

Leave a Reply

Explore our other AI Insights or Recent Posts.