The Center for Hybrid Intelligence (CHI) has with partners such as UN’s AI4Good and the Danish People Climate Summit piloted a new AI powered game, “Crea.visions,” which allows the general public to collaboratively create and reflect on images of possible futures. Such accessible crowdsourcing of the concerns and solutions of the general public on both global and local complex socio-scientific problems could form a cornerstone in participatory democracy in the digital age.
With the emergence of global societal dilemmas, large scale public participation in the formulation and implementation of a wide range of local and global solutions is more important than ever. Yet, in many ways interest and perceived agency for such participation is waning rather than increasing, posing both a challenge to democracy and the globe. In response to this we created Crea.visions. an Artificial Intelligence (AI) supported game designed to let anyone in the general public blend images of a possible future.
The objective of the first 5 studies with this game and supplementary workshops is three fold
1) Lower the barrier for the general public in contributing to the dialogue on complex societal dilemmas.
2) help residents of cities around the world to envision the future of their city through AI powered image blending.
3) Demonstrate that custom AI models can be trained to adjust to local scenarios (crea.visions.Paris and crea.visions.Venice)
Crea.visions was first developed in 2020 with the United Nations AI for Good and ArtBreeder in order to raise awareness about the climate change challenge and the importance of reaching the UN’s Sustainable Development Goals (SDGs) (https://www.youtube.com/watch?v=SpFN7blPvhY). Since the launch of crea.visions, versions of the game have been successfully used in Denmark at the Danish Peoples Climate Summit in Fall 2021, in Middelfart, Denmark, in April 2022 with the Steno Science Museum in Aarhus, Denmark in May 2022 in Venice, Italy as part of ACM’s Creativity and Cognition Conference (https://cc.acm.org/2022/crea-visions/). Additionally, crea.visions will be used in a series of events in January 2023 as part of the international Learning Planet Festival in Paris (https://festival.learning-planet.org/).
In our work we combine citizen science methods (e.g. when the public voluntarily helps conduct scientific research) with participatory democracy (individuals influencing policy directly, not through voting) and human-AI co-creativity (humans and algorithms working together in a creative process). It involves multi-stake holder alignment and participatory design.
The first pilot with AI4Good was launched entirely virtually and resulted in 1624 submission. The remaining pilots have been released in connection with physical events. The Climate Summit version was played by +500 high school students at a Youths Summit as well as at the physical summit resulting in 237 submissions and prize ceremony hosted by the City Mayor in connection with the conclusion of the Summit. Crea.visions.Venice was organized with the Creativity and Cognition Conference and included a series of pop-up workshops in schools, bars and public venues and culminated in an art exhibition of the 16 most expressive submissions. This art exhibit will now travel world-wide with next planned appearance in Edinburgh in 2023 in connection with the Edinburgh Science Festival. Crea.visions.Paris takes the innovation even further by building in explicit gateways of civic action in collaboration with a selection of local NGO’s.
We were inspired by the online community around the project Artbreeder, in which visitors use the AI-trained models to generate and iterate on digital art generation. This has resulted in an amazing breadth of contributions some of which have been sold commercially. Our main innovation has been to adapt the user interaction to collective addressing societal issues.
What Makes Your Project Innovative?
AI technologies have been developing at an amazing pace over the past decade. Perhaps one of the most striking examples is the emerging field of computational creativity in which algorithms autonomously generate human-level “creative” products such as visual art, poetry and music. This trend is most prominently seen in the explosion of research and technological outputs on text-to-x (image, video, etc) spearheaded by the release of Open AI’s Dall-E in January 2021 (OpenAI, 2022).
However, to-date, no one has specifically used these technologies to help the general public create solutions to local or global problems that they or their community is facing. Combining these cutting edge technologies with appropriate scaffolding has the power to change the way the public engages in dialogs about complex socio scientific problems.
What is the current status of your innovation?
The project has been deployed in several real-world settings and been subject to scientific study and validation and as such is ready for deployment and scale-up. However, with every new version of crea.visions player interactions are continually refined and features added supporting more intuitive interaction with the AI and deeper engagement in the particular dilemma. As an example, within the first game version, submissions consisted only of either a utopian or a dystopian image, whereas the crea.visions. Paris version will feature title, text identifying both the identified problem and the proposed solution as well as geotagging, discussions fora and links to further community action possibilities.
Collaborations & Partnerships
Citizens were involved in the every design iteration of crea.visions.
AI4Good (https://aiforgood.itu.int/) helped facilitate the first public launch.
The Mayor of Middelfart in Denmark and the Danish Peoples Climate summit helped facilitate the 2021 event
The Creativity and Cognition Conference helped facilitate the 2022 event in Venice.
The Learning Planet Institute will help facilitate the 2023 event in Paris.
The Carlsberg and Novo Nordisk foundations financially supported the project
Users, Stakeholders & Beneficiaries
- Citizens were able to contribute their thoughts on the future
- Government officials are able to see what citizens/residents are interested in
- Civil organizations have an innovative means of engaging the pubic in their cause
- Researchers are able to study how co-creative tools can help support participatory democracy
Results, Outcomes & Impacts
In total, over the five events, more than 25,000 images have been produced.
Scientific studies of crea.visions have documented enjoyable in-game experiences and innovative and artistic reflections about city issues (e.g. climate change, tourist, deforestation). For example, the initial results from the 2020 launch were published in the Creativity and Cognition Conference
https://dl.acm.org/doi/fullHtml/10.1145/3450741.3466815. The main findings of these preliminary results show that crea.visions can facilitate a wide range of utopian and dystopian images and the creators intent of the images are recognizable by viewers.
The focus of the crea.visions event in Paris is to explicitly support pathways for sustained civic engagement from the game interaction and into the real-world. For this we are partnering with local NGOS in Paris facilitated through the Learning Planet Institute.
After Paris, emphasis will be to deploy in several communities at once.
Challenges and Failures
Working at the state-of-the-art of AI-enhanced creativity support and massive digital civic engagement entails a large number of challenges. Technically, the AI needs to be trained with substantial amounts of visual material from the local context in order to smoothly generate locally relatable outputs. Designing intuitive ways of interacting with such AI systems is a huge outstanding challenge for us and the entire field and can only be addressed with iterative development and field testing.
Having established rewarding human-computer interaction solutions the interdisciplinary challenge of embedding this interaction into a meaningful social context had to be solved. Again, there are no standard off-the-shelf solutions and continual experimentation is needed. We have experimented with both monetary, extrinsic motivations in the form of prizes as well as intrinsic motivation stressing the opportunity of the participants to get a public voice in documenting their concerns and ideas.
Conditions for Success
At an individual level, all previous versions of the game have been an immense success documented by the overwhelmingly positive reactions and testimonials from all involved stakeholders. As the project transitions from pilot stage to full deployment more tangible success criteria documenting value-for-money need to be developed. This will involve documentation of sustained increased civic engagement of the participants and documentation of tangible civic suggestions and dialogues supported by the game interface.
This will require additional formal buy-in from the local, regional and global governing bodies to sanction such large scale efforts and welcome the resulting community inputs. For a truly large-scale deployment additional financial resources will required both for development and deployment support as well as technically for additional server capacity.
We have so far conducted 5 events with crea.visions in various capacities, always implementing new features and designing the game to be context specific. We are currently in the planning phases of crea.visions.Paris in January 2023 and crea.visions.Aarhus later on in 2023.
We hope to eventually partner with large organizations such as the UN to create large scale events that reach diverse and traditionally underserved communities
We welcome collaborators to co-create crea.visions events with us.
Working with state-of-the-art AI has been an immense motivator and technological challenge. The degree of technical expertise to develop and support such an activity should not be underrated. At the same time, the potential communicative power of utilizing AI for good is immense. Finding a proper way of communicating this “hybrid intelligence” interaction in an understandable way to the participants and other stakeholders has proven immensely rewarding for the project.
Another valuable lesson has been the immense eagerness of citizens from all ages and parts of the society to engage in these gamified activities. Working with eg. retired citizens in their 80’ies and observing them actively engage in the AI-interface and the surrounding real-world community discussion has been immensely rewarding.
- Identifying or Discovering Problems or Opportunities - learning where and how an innovative response is needed
- Generating Ideas or Designing Solutions - finding and filtering ideas to respond to the problem or opportunity
- Developing Proposals - turning ideas into business cases that can be assessed and acted on
- Implementation - making the innovation happen
- Evaluation - understanding whether the innovative initiative has delivered what was needed
- Diffusing Lessons - using what was learnt to inform other projects and understanding how the innovation can be applied in other ways
16 January 2023