Using AI before a participatory process starts
/Using AI before a participatory process starts
Streamline administrative work
In addition to the AI features offered by participation platforms, which we'll discuss in detail here, participation hosts are using AI for discrete back-office tasks related to running a process. In these cases, they might not use a dedicated participation tool, but a general purpose AI model. They're using AI to do things like:
Editing dense legalese and government or corporate jargon to make them more accessible to regular people
Digitizing analog records from archives and offline communication channels
Streamline communications, such as scheduling
Dipping your toes in
Introducing AI can feel overwhelming. What's the 101 version? Where can you get a good return on minimal investment? After closely studying AI applications, the Civic AI Observatory program by Nesta and Newspeak House wrote up some likely "safe bet" parameters for public good organizations looking to try out AI. What kind of task might be a good candidate for automating with AI? Edward Saperia suggests starting with basic tasks that are:
Done frequently
Supervised internally
Involve limited downsides if they go wrong
Can be easily compared against an existing process
Require limited technical investment
Similar to tasks that have been automated successfully by other organizations
Not better solved by a tool designed for that task
What about bigger, riskier projects? Saperia says, "It seems like a good idea to start with the basics, and also — given the speed of change — wait for the private sector to make all the mistakes first, and also develop the tooling.”
Plan your process
Example platforms with this feature: IdeaScale, deliberAIde.
AI tools can help participatory process hosts create and plan their actual program.
For example, IdeaScale AI provides users guidance and recommendations to design their innovation campaign. Likewise deliberAIde, as seen in this screenshot provided by the company:
Brazil's Empurrando Juntas recommends the use of AI for this purpose, but does not offer a feature to facilitate it on the platform. Participatory process hosts are also using mainstream models like ChatGPT and Anthropic to brainstorm and develop participatory programs, although you should provide the AI with plenty of background context on your program if you go this route.
Training staff to run participation processes
AI can be used to dynamically generate documentation for a platform that doesn't have sufficient guidance. It can also create videos, slide decks, and other engaging materials to help train staff to moderate the participatory process, for example.
Swappable AI models
Example platforms with this feature: Your Priorities, Sensemaker, Decidim
Some participatory platforms allow administrators to choose from a variety of third-party AI models to integrate. This provides admins the ability to choose a model more in line with their values, procurement rules, or preferred languages. It also allows them to swap in public interest AI models in the future once they become available. This offers a compromise between using today's advanced models and migrating to public interest models in the future when they come online.
Dynamic FAQ and Glossary
Example platforms with this feature: StaffLink, Deliberaide (on the roadmap)
Before starting your participatory process, you can prepare dynamically generated Frequently Asked Questions or Glossary sections. Rather than (just) pre-write answers to the questions you expect people to have, you can aggregate common questions to your program's chatbot and publish answers to them. POPVOX Foundation did this with their StaffLink tool. With AI, their system "groups questions by topic, generates a generic version of the question, and shares the number of times a related question was asked."
POPVOX's Frequently Asked Questions section dynamically generates generic versions of the most common questions its users have actually asked its service. https://stafflink.popvox.com/faq
Smart groups: Form diverse participant groups
Example platforms with this feature: Go Vocal, Unanimous AI, Frankly, Your Priorities
Forming diverse, balanced, and representative participant groups is key to the success and political legitimacy of many participatory democracy processes. Platforms are using AI to dynamically create and balance participant groups. For example, the AI can take local demographic data into account to populate groups with representation from key demographics.
Simulating participation to test your platform
Some platform developers are using AI as a form of pre-launch user testing to simulate people joining and using their websites. This can help you determine where in a sign-up process people might get discouraged and drop out, for example. Many platforms invest time and energy into optimizing the first 'landing page' that visitors reach, for example, to maximize the number of people who ultimately sign up.