Chatbots Against Climate Change
This article is a summary of my master thesis project in interaction technology & design at Umeå University. My study analysed the use of chatbots as a mean to motivate people to live more sustainable lives. In this article you’ll find my learnings, experiences, and thoughts on designing conversational interfaces as a mean to motivate behavior change.
In January 2017 I started doing my master thesis in collaboration with the digital agency Daresay. We decided that my thesis project would address the global goals of sustainable development and modern technology in some way. I dived down into the global goals and several articles to quickly realize that I wanted to work with goal number 13, climate action. Partly because the subject is wide with great potential and many approaches, but also because climate change is something that is constantly being discussed (yes, believe it or not we still have world leaders denying it 😓).
So, I started at looking on what had been done in terms of technology and climate change. I noted that the biggest problem with climate change, is us. To reach a sustainable future we all need to take urgent action to combat climate change and its impacts. This is where I knew I wanted to get a better understanding on how to motivate behavior change with conversational user interfaces (CUIs). CUIs are discussed as the future of interfaces, but can they be used to persuade and motivate our everyday behavior?
To find out, I began by reading a bunch of articles. Articles mostly related to behavioral science, motivation and CUIs. The following sections is a summarized version of what I found.
Motivation is commonly known as the driving force that enables certain behaviors. It can be defined as people’s direction to behavior and is many times the reason for our actions, desires and needs.
“Motivation can be defined as people’s direction to behavior.”
Stephen Intille suggests that there are five components of presenting messages to motivate behavior change effectively:
- Present a simple, tailored message that is easy to understand.
- The message should be presented at an appropriate time.
- The message should be presented at an appropriate place.
- It should also be presented using a non-irritating, engaging and tailored strategy.
- It should be presented repeatedly and consistently.
He also concludes that presenting information repeatedly and consistently may be the greatest ubiquitous computing challenge. To prevent a message becoming annoying is to ensure that the message has a high value for its user and that the message does not appear judgemental.
There are many motivational factors that affect pro-environmental behavior. This is some of the most commonly used motivational factors that I stumbled upon during my research:
This motivational factor says that climate change and its impacts is a problem that requires signaling, illustrating, and explaining. Knowledge of issues and of action strategies can inspire people to be more environmentally friendly.
During my research I also found out that goal-setting is a well studied source of motivation. Setting goals to motivate people has many functions; they serve a directive function, they are energizing, they motivate us to do more, they affect persistence, and they affect our behavior indirectly.
We are highly motivated by one another aswell, and our actions are strongly shaped by the people surrounding us. If the surrounding culture and people propagates a sustainable lifestyle, pro-environmental behavior is more likely to occur.
A commitment can be defined as a pledge or promise to behave in a specific way or attain a certain goal. Studies show that a person that expresses commitment towards a certain goal is more likely to pursue that behavior.
Incentives & Rewards
Incentives and rewards are also common motivational factors. Incentives and disincentives are antecedent motivation techniques that come before a certain behavior, and rewards and penalties are consequence motivation techniques that come after. Incentives and rewards does not always have to be economical; status or convenience may also have important effects on pro-environmental behavior.
Chatbots & CUIs
Before discussing CUIs and chatbots a brief definition of conversation is needed. In the Oxford English Dictionary I found this definition, which I believe encapsulates all different aspects of a conversation:
“A conversation is a talk, especially an informal one, between two or more people, in which news and ideas are exchanged.”
CUIs enable people to interact with smart devices using spoken language. This can be through a chat, through plain text or speech. CUIs has been around for many years, but it is in the recent years that is has got increased attention. Various technological advances have contributed to the rise of conversational interfaces. Technologies such as Artificial Intelligence, speech recognition accuracy which has followed the adoption of deep learning.
Messaging platforms are today increasing in numbers and the way we communicate and exchange information is with mostly through text messages. This combined with the increased interest for CUIs has led to a buzz for chatbots as well and in April 2016, Facebook released their own chatbot feature in Messenger. The main purpose was to increase people’s experience with the platform and to let businesses reach out to their customers in a completely new way. To make it easier for developers and designers to build beautiful and consistent messenger bots that allows for a unified experience, Facebook released design their own guidelines to follow.
Designing & Developing Chatbots
Once I had gathered theory about behavioral science, motivation and chatbots I decided that I wanted to create 3 chatbot prototypes based on the motivational factors mentioned above; one informative, one goal-setting, and one comparative. I also decided that the chatbot prototypes were going to be developed in order to change people’s food consumption habits. To design the conversational flows I used the following tools:
- Pen and paper
Twinery is an online tool that gives a clear overview of possible outcomes, user behaviors, and user needs of the conversations. For prototyping I used tools suggested by Bashmakov to get the chatbot prototypes up and running quickly:
- Chatfuel — A graphical programming language that makes it easy for anyone to host a messenger bot.
- Glitch — A NodeJS environment hosted in a cloud where users can edit each file online.
- QnAMaker — Natural Language Processing platform for more complex interactions.
The prototypes were then developed using an iterative design process in 4 main steps; analyze, design, prototype and evaluation. The user tests were conducted on 7 people per prototype, all participants used the chatbot for one week. The duration of the tests was chosen because I wanted the users to get a feel for the personality of the chatbot and their thoughts on how the information was communicated by the bot. I also wanted to see if the users saw any difference in their food consumption behavior and if the chatbot actually had affected them in any way. My findings are presented and summarized as two separate guidelines in the following sections. One guideline is proposing a way to design for chatbot interfaces, and another one that describe my suggestions on what to consider when designing for motivation in conversational contexts.
Design Principles for Chatbot Interfaces
Getting started with designing for conversational user interfaces can be a bit overwhelming. What is the best way to design for chatbot interfaces? Where do you start, how do you start and in what way should a prototype be iterated? I propose a seven step process to design for chatbot interfaces:
- Define Goals
I believe that, as for any other service, the goals and intents of the chatbot and users has to be defined at an early stage in the project. People are going to use the chatbot for a particular reason, you should pinpoint these reasons and let them be a groundwork for the continuation of the development process.
When the goals of the chatbot has been defined I would suggest that you write down a list of keywords that corresponds to the goals of the users and the chatbot. These are going to be of great help when creating the conversational flow and the personality of the chatbot in the following steps.
Early on in the design process decide which platform the chatbot is going to be hosted at. This step is more important than you would think. Developing chatbots for different platforms can differ significantly when it comes to interaction possibilities, input-styles, placing of text, and so on. The chatbots goals and the user intents are key factors of choosing the right platform. For instance, if the chatbot needs to reach its user at any time of the day, I would suggest using Messenger. But if the chatbot is intended to reach its users at work, Slack could be the better choice.
“The chatbots goals and the user intents are key factors of choosing the right platform.”
4. Simulate Chatbot with Wizard of Oz
Based on the goals and the keywords from previous steps, I would highly suggest that you simulate the chatbot on the chosen platform. Since chatbots are conversational, what better way is there to define the interactions than through a real conversation? A suggestion is to use the Wizard of Oz method. Wizard of Oz is a rapid prototyping method where a person simulates a system’s functionality and interacts with a test person through the interface. This will give you a good idea on what people want to talk about to your chatbot and a great way of getting started with the design of the conversational flows.
5. Conversational Flows
Once you have found some key conversational triggers the conversational flows can be designed. A great way to start is with a text-editor to write down potential flows. When the conversational scenarios are developing to more complex ones, use Twinery go get a clearer overview of how it all comes together.
6. Chatbot Personality
Once the conversational flow of the chatbots is designed, you should create the chatbot personality. This step is very important because the graphical appearance and the tone of the chatbot is going to reflect its personality.
7. Write Scripts
Last but not least, the chatbots script can be written. Facebook’s design guidelines are good to follow. Key scenarios to have in mind here is the onboarding process, to always have a fallback script, to have reminders, and to be clear about the chatbots features and functionalities at all times.
“Key scenarios when writing scripts: onboarding, fallback, reminders and to always be clear about features and functionalities.”
Bring it all together
Once your done with the scripts, the chatbot has a steady ground to be built upon. It has a personality, a tone, a graphical structure, a home in the shape of a platform, and it knows how to answer user input. Now your bot is ready to talk to its target group. Be ready to constantly change and improve the bots structure and conversational skills.
Motivate with Chatbot Interfaces
The main goal of my thesis was to present guidelines on how to motivate pro-environmental behavior with chatbot interfaces. Throughout the thesis I’ve compiled 8 aspects to keep in consideration when trying to motivate behavior change with chatbot interfaces;
- Catch Interest
- Model Conversation After Users Preferences
- Positive Personality
- Graphical Appearance
- Spontaneous Messages
- Responsibility & Reciprocity
- Trigger Points.
Catching user interest is an important factor to think about when designing for pro-environmental behavior. The primary goal is to show users how they can affect the climate situation. To do so, pinpointing activities where people need to make a change makes it easier for them to relate to the problem. It is important that the facts are communicated in other ways than pure text aswell; it needs to be illustrated, displayed, and explained in an intuitive way that people understand and catches their attention.
Model Conversation After Users Preferences
To affect people’s pro-environmental behavior with chatbot interfaces, modeling of the user is required. It was obvious from my user interviews that people’s perception of the chatbots varied significantly. It is important that the chatbot learns its user’s personality in order to communicate sustainable information in a way that suits them. Fogg show that similarity is one of the most powerful persuasion principles. Conversational systems that are similar to the people using them motivate and persuade them more.
Personality is crucial when developing for a pro-environmental chatbot. From my results it was obvious that people felt at ease when talking to the positive chatbot, despite the severity of the climate situation. Information should be communicated in a positive manner by a positive being. It is more likely to drive individuals to take action that benefits the climate with positive and good emotions, rather than with negative ones.
Another important aspect of personality, language and motivation when it comes to behavior change, is to use praise. My study show that users felt good about themselves when they were praised. Praise from a system or a computer can generate positive effects similar to the one from people, which in turn can motivate behavior change.
“Praise from a system or computer can generate positive effects similar to the one from people.”
It is important that your chatbots graphical appearance coincides with its personality. People are often drawn to things that they can relate to and for technology to have some kind of physical characteristic can be enough to convey social presence. Attractive technology is believed to have an attractive force, people tend to like applications or services that they think are beautiful. Research has also shown that it is easier to like, believe and follow attractive people. This should therefore be applied to chatbot interfaces as well, in order to motivate and affect people’s behaviors.
Design your chatbot so that it is spontaneous in its way of messaging as well. From my user tests, people discussed how much more natural the chatbot felt when it talked about subjects unrelatable to climate change. Unrelatable messages and psychological cues infer that the chatbot has emotion, and emotions can be used to motivate and persuade users to change behavior.
Responsibility & Reciprocity
From my tests some people felt a responsibility towards the chatbot. This is interesting because it shows the social impact that chatbot interfaces can have. Some people felt more responsible towards the chatbot then they did to the climate. The reason for this can be of the social rule of reciprocity, which states that after you receive a favor you must pay it back. A great way to create reciprocity is through a kind tone and repetitive messages, in this way people might feel responsible towards the chatbot.
Trigger points was shown to have major impacts on the users. It is important to find features and functionality that is close to the goals of the chatbot and the behavior it is going to motivate.
When developing for chatbot interfaces designers have to rethink their process. Designing for Android, iOS or Web means that there is a lot of focus on style, fonts and colors. But in a chatbot interface these do not need to be taken into consideration at the same degree. The content is the style of the chatbot and the focus should be on how this content is communicated; with the personality and appearance of the chatbot. When testing a chatbot interface it is important that users have the opportunity to get acquainted with the system in different ways. A chatbot has its own personality and just as with new people it takes a while to get to know them.
My study could not show any statistical evidence that people’s pro-environmental behavior can be increased with conversational user interfaces or chatbots. However, the result from the user interviews indicates that chatbots can affect and motivate people to consume food in a more sustainable way. We all need to change behavior in order to reach a sustainable future. To find the right way to do so, different digital channels need to be tested. Social influence has shown to have a big impact on people, if a chatbot can get into this social sphere it has the ability to motivate people’s behavior on a completely different level than any other form of technology.