What is the “autonomous world?” As Jeremiah Owyang, founder of Crowd Companies, describes it as drones, robots, chat bots, and AI. He should know, having tracked and observed the disruption caused by online technology over the past decade.
We are in the middle of global disruption due to widespread mobile Internet and cloud technology, Big Data, the rise of the collaborative economy and crowdsourcing/funding. As Jeremiah points out, we are entering the next phase where robots, drones, bots, AI, and self-driving cars will have widespread adoption.
And after that technology will be merged or implanted in humans. Think about swallowing a micro medical device that can send a report back to your doctor or having your fitbit implanted in your body so you don’t loose it. Think cyborgs
What does this mean for the nonprofit sector in the future? I had an opportunity to walk the Stanford Dish recently with Jeremiah the other day. And one thing he said about the impact of living in an autonomous world, “Empathy, human touch and interaction will be become more important.”
It reminded me of a prediction that my TechSoup Global colleague, Marnie Webb, said to me almost 7 years ago when we were talking about the impact of crowdfunding and connectivity. She said that in a world of connectivity where individuals can do their own crowdfunding for services, bypassing nonprofits, that nonprofits will still have the advantage because of the human relationships.
A recent post on the Ford Foundation blog, “Why you should care about bots if you care about social justice,” notes automation itself isn’t cutting edge, but the prevalence and sophistication of the how the automated tools interact with users is. And, some can spread beauty and inspiring art, like the MoMaBot or abuse and misinformation instead.
NGOs, activists, and even governments have used bots to automate positive social change activities:
The post offers some recommendations to nonprofits and activists about how to avoid having bots (evil bots) disrupt their activism and creative ways they should be thinking about deploying automation in service of social justice, especially if the nature of their work might make them likely targets.
- The potential for an army of bot trolls to disrupt an activism campaign
- A policy for how to respond to negative comments or social media workflow
- Bots could help protect the privacy and safety of activists
- Bots needs to be designed increase engagement, support and inspire offline action
Nonprofits are also using bots for supporter engagement, fundraising, and cultivation and as part of their digital strategies. Right now it is difficult to discover bots, but after a bit of research I found a few examples of nonprofits making use of Facebook Messenger platform.
- National Geographic the Albert Einstein Bot to accompany its Genius Film Festival. This is one of the best examples I saw of using bots for engagement and it most definitely required some custom programming.
- Climate Reality is using a bot on its Facebook Brand Page. The bot is programmed to encourage fans who message the page to sign up for different actions. It also reminds folks that it is robot. This bot is for lead generation and was created using one of the free chat creation platforms for messenger which I discuss below.
- Charity:Water created a bot called “Walk with Yeshi” as a donor engagement tool. When you send a message to this Facebook Messenger bot named Yeshi, she tells her story about how her daily walk for clean water in Ethiopia.
- The UN Global Report is using a bot to engage young people in discussions about civil society related to the reports.
I pulled together a curated list.ly of some really well-designed bots from nonprofits and beyond here.
Designing an Effective Nonprofit Chatbot
I decided to set up an experiment to better understand what was involved in the design of small pilot bot. Since I am not a coder, I used Octane AI, an app that let’s anyone create an app co-founded by Matt Schlicht and Ben Parr, to set up “Beth Bot,” for my Facebook Brand Page.
To design an effective bot, you first need to figure its purpose and how it connects to your goals. You also need to think about its voice and tone and how it can be relevant and engaging to your target audience.
There are two design tasks. First, if you are not doing custom coding with NLP like the Einstein Bot, bots work best with close-choice questions that lead the user to more information or an action. The Whisky Bot, while not a nonprofit focus, it is a good example of a more elaborate bot with close-ended questions guiding the user. The Climate Reality bot is a good example of a simple bot to build an email list.
The next step is figure out the script and pathways for the conversation. Bots can also answer open-ended questions and get smarter as they go, if you train them. That part is an interactive process where you add answers that users ask the bot. Because Bots are somewhat of a novelty now, users seem to enjoy trying to stump the bot with open-ended questions. Using a platform to create your bot where you don’t need to know programming language, does not yet let you create a bot that can chat with intelligence.
And that brings us to the another task, testing and iteration. I started there by asking colleagues to ask open-ended questions so I could train the bot. When people asked it questions that it doesn’t understand, you program it to respond with a selection of responses. (On the back-end, you can go in add responses to those questions so the next time someone asks that questions there will be response. ) Right now the software is super limited on the training part, but as the platforms evolve this will change. (I hope)
Training your bot is an iterative design process that is required for a bot created without having to put your hands into the messenger code and if you want to make it engaging. There are bots that are self-training using AI language.
The testing was fun and it also helped me figure out how to script the close-end conversation that happens when you “Wake Up the Beth Bot.” The app was fairly simple to use, but it was really helpful for me to sketch out the flow first and then the conversation based on the pathways the user could take. It is helpful to keep your responses short and incorporate emojis, images, and even video.
If you want to see how the Beth bot carries out a conversation and responds to open-ended questions, you can engage with the Beth bot here. The app gives you a dashboard that gives you metrics and you can also track conversion rates on the landing pages that you send users to.
If you want a basic primer on what a chatbot is, I suggest this article from Chatbot Magazine. To program a bot in Facebook Messenger, you don’t necessarily need deep technical skills as there are apps that make it easy, including Octiveai, Manychat, or Chatfuel.
Here’s a few good tip sheets if you want experiment.
- The Complete Guide To Creating a Facebook Messenger Bot
- Beginner Guide To Growing Your Audience on Facebook Messenger
Bots are catching on and as we move further into the autonomous, your audience/stakeholders may expect them as a way to connect with your nonprofit. The most effective bots are those that engage, but that requires some investment of design and iterative development. One easy way to get started is to create a pilot with a specific goal, measure it, and develop it further based on what you learn.
Is your nonprofit experimenting with or thinking about using chat bots? Please share in the comments.