Mary Stevens23 Aug 2023
What if… the most divisive issues we face in our communities could be where we find common purpose?
What if… debates around traffic reduction and air quality could become the place where together we start to imagine cities that allow everyone to thrive, starting with the most vulnerable?
What if… our experiences of exploring these issues online allowed us to understand our neighbours better, helped us to appreciate new perspectives and find common ground?
And what if it were AI that helped foster this shift towards collaboration?
Conflict or consensus
This might feel far-fetched. For now, conflict sells. The AI-based algorithms that determine what we see in our social media feeds have learned that the most extreme views are what keep us scrolling (and more scrolling means more ads and so more profit). But just as AIs can be directed to maximise profit through polarization, so they could be re-directed, if we choose, to optimize for empathy and consensus. And that empathy could even extend beyond humans - to the rest of the living world.
AI and active citizenship
The threat that Generative AI presents to democracy is increasingly well-documented, and in the context of environmental action the risk of accelerated automated climate mis- and disinformation is ‘tremendous’. However, in our recent lab we also identified the opportunity to apply responsible AI to support active citizenship. We cannot and should not just accept as given that AI can only be deployed in the service of greed, at the expense of our institutions, and our living world. Alongside regulation and frameworks to curb the harms, we can also bring to life alternative possibilities.
There are three key areas where we think AI could enhance active citizenship
More inclusive and engaging consultation. Tools like pol.is already seek to highlight consensus rather than difference, focusing on giving the participants the opportunity to propose constructively, rather than undermine through objections. (It’s not technically an AI tool, but it’s evolving in that direction).
Cortico.ai is another tool, only used in the United States so far. It's focused on improving public dialogue, creating more inclusive civic spaces and supporting decision-makers to listen better - all backed by the power of large language models. In the future AI assistants for public and online dialogue could
- highlight the voices that are missing, anticipate likely sticking points
- act as a coach or prompt in contentious conversations to support people to find common ground,
- develop engaging scenarios to explore the consequences of different choices and even represent the perspectives of future generations or other living beings (how do the bats or the birds feel about a proposed development?).
When AI was used in a recent study in the US to propose more constructive online comments on the topic of gun control to participants with strongly opposing views, researchers found that the AI assistant “improved the reported quality of the conversation” and “reduced political divisiveness.” What if we could do the same for phasing out fossil fuels?
Democratising access to specialisms. As an ordinary citizen, anyone who has ever tried to respond to a planning application or influence a strategic planning document, like a local development plan, will know just how bewildering the process is. It’s hard to navigate the different stages of an application or understand what constitutes a legitimate (‘material’) objection. Finding the information that really matters in all the documentation can also be very hard.
At Friends of the Earth we receive many more requests for help responding to planning applications than we can possibly support. DemocracyNext has recently established an impressive international panel to look at democratizing urban planning. But what if an AI assistant could act as an expert planning ‘co-pilot’, reading documents to identify the areas where plans are inconsistent with our priorities and vision and supporting people to submit impactful statements?
Building community power. Campaigning is all about influencing decision-makers. We do this by writing letters, arranging meetings, speaking to people in power about the things that matter to them – and to us – and trying to persuade them to act in the interest of our cause. The potential of AI to learn what works for people, and to create personalized, targeted narratives can work for everyone. It can work for vested interests and people trying to sell us things we didn’t know we wanted – but it could also be used to support campaigners to understand the concerns of their decision-makers and how best to speak to them.
It could coach and train activists in having impactful face-to-face conversations, by pre-empting the objections that are likely to come up and suggesting compelling counter arguments. We first developed a storyboard of this idea in 2018 (the ‘Activistbot’), but at that time the data simply wasn’t available to train a model. Easy access Generative AI has removed this objection.
A transparent and ethical approach
In all these areas it will be important to adhere to clear ethical standards and to be transparent about how we are using AI and what for. The cost-benefit analysis will have to stack up too. Is the outcome – a better letter, a better conversation – worth the energy, water and raw materials that will have to go into creating it using large language models? In all things we need to pay attention to the extent to which AI is enhancing rather than undermining our uniquely human capabilities. Are the tools we use helping us reclaim our attention and focus, or further capturing them?
We would like to start prototyping these ideas and gathering real-world feedback. But we don’t have the resources or the capacity to do this alone. We are sharing these ideas in order to understand what resonates with people.
Are you already working on a related project and would like to think about how we might collaborate to secure funding?
Are you a funder wanting to understand how to align these technologies with your goals?
Are you an academic curious about real-world applications (and able to access funding to support your research)?
This is the second blog post from a series of four, exploring the outputs from our recent design lab. Mary Stevens looks at opportunities for active citizenship. If you are interested in collaborating on any of these ideas please get in touch at [email protected]