Collective intelligence - people working together, with the help of technology, to mobilise a wider range of information, ideas and insights - has played a key role in responding to the coronavirus pandemic. It has demonstrated the power of distributed intelligence to meet urgent data, knowledge and skills gaps, and the ability of the crowd to organise for pandemic response in a more devolved way.

In September 2020, the Centre for Collective Intelligence Design launched the third round of the Collective Intelligence Grants programme. We offered funding grants of up to £30,000 to experiments that offer innovative ideas that could help create more equitable and sustainable futures.

Between April and December 2021, we will be supporting 3 diverse organisations to conduct the following experiments:

Who is behind the experiment?

DigVentures in partnership with ArchAI and Bright Water Landscape Partnership.

What is the experiment about?

This experiment will test a new collective intelligence approach to identify previously unknown archaeological and historic features on mapping data of a place.

It will test if the combination of crowd-based labelling of Geographic Information System (GIS) and LIDAR data can be used to train an AI that can identify sites of archeological interest. It will explore how this can be used to inform local planning decisions, whilst making local people central to the process.

How will this experiment add value?

The experiment will be an opportunity to co-create datasets that underpin and describe the historic environment, preserve the distinctiveness of places, protect fragile archaeology and position local people as stakeholders in the planning process. This methodology has wide implications for public and private land transformation projects, from climate change and carbon capture to housing development and large-scale infrastructure.

This is the first time Collective intelligence and AI will have been deployed in spatial and participatory planning.

How might the findings help people better design collective intelligence?

The experiment will demonstrate how crowds and AI can work together in ways that could be applied to other landform and object types (apart from previously unknown historic features). It will also demonstrate how to train crowds to work with novel remote sensing datasets and how to use digital methods to support engagement.

Who is behind the experiment?

Grantee University of Manchester in partnership with E.Mission Innovations Ltd and Wythenshawe Community Housing Group: Real Food Wythenshawe Project.

What is the experiment about?

This experiment will test whether a natural language processing (NLP) model trained by a crowd can be used to calculate the carbon footprint of online recipes and suggest lower carbon substitutes. The experiment will also test if presenting people with alternatives generated by the model helps to facilitate lower-carbon cooking.

How will this experiment add value?

An individual’s diet comprises approximately 25% of their carbon footprint. Since the outbreak of Covid-19, home-dining has increased by around 10% in the UK, presenting an opportunity to raise people’s awareness of the carbon footprint of their home-cooking.

This experiment aims to reduce the domestic food emissions by making data on food emissions more accessible, and offering crowdsourced acceptable substitutes.

The NLP-aided carbon calculator will be made open source at the end of the project.

How might the findings help people better design collective intelligence?

This experiment will contribute to expanding understanding of how NLP and crowdsourced content can be combined to promote behaviours and practices that contribute to net zero targets. The experiment provides a novel use of crowdsourcing in the context of sustainable cooking practices.

Who is behind the experiment?

King’s College in partnership with Google (NY)

What is the experiment about?

This experiment will test public trust and confidence in an Artificial Intelligence (AI)-enabled system that rates the readability of different charts used to communicate current and emerging issues surrounding the UK’s post-pandemic recovery.

The team will use crowd collective intelligence (CI) to train and test machine learning models that identify misleading or inaccessible information in popular chart types. It will also assess how much people are prepared to trust the AI-generated rankings.

How will this experiment add value?

This experiment will contribute to understanding how people engage with charts and how specific chart-design elements impact on their ability to understand what a chart is about and trust what it is trying to convey. This has direct implications on the design of data stories and how information can be best presented to the public. The experiment will map the process of collecting and using CI generated input to generate AI outputs/rankings when designing public information, which is a powerful workflow to support scientists, analysts, journalists and policy makers in creating more accessible and understandable ways to communicate important data.

How might the findings help people better design collective intelligence?

The experiment will expand our understanding of how people react to AI-generated rankings and how they impact people’s judgments. This has particular implications for AI supported fact-checking services. It will also produce insights and guidance for policy makers, scientists and journalists on how to better communicate complex data.

Read more about the second round of the Collective Intelligence Grants featuring 15 grantees who conducted new experiments in machine-crowd cooperation to address complex challenges (launched in September 2029). The findings will be published in summer 2021.

Read more about the first round of grantees and the 12 diverse experiments supported between March 2019 and January 2020, including their results which have been published in our report ‘Combining crowds and machines’.