AI Collaboration to Improve Quality of Life
UNICEF’s CryptoFund invests in two Latin American companies for a joint open source project with a local branch of the Argentine Government
In another first for UNICEF Ventures, two of our portfolio companies have joined efforts to deploy a project funded by our CryptoFund in partnership with the government of Cordoba, Argentina’s second largest city.
Atix Labs is an Argentine company specialized in building digital products that require blockchain or artificial intelligence (AI) implementation. During the last 5 years, they have successfully implemented solutions on Decentralized Identity, Asset Tokenization, DeFi, DAOs, Digital Signature, Transparency Platforms, Stablecoins, Image Recognition, and Machine Learning — most of them as the first implementation of its type in Latin America.
OS City is a Mexican/Argentine govtech company offering Open Source digital identity and certification solutions using blockchain and AI to increase transparency, preserve data integrity and accelerate the use of portable records in the public sector.
The joint project consisted of building an open-source platform to easily understand, classify and analyse citizens' concerns and expectations about the use of public spaces in an Argentine city, from an accessibility and inclusion perspective. Such analysis was enabled through AI and will allow the Argentine government to save time reading thousands of public reviews in order to act more quickly and efficiently to resolve any discontentment or meet any preference from the public. The project was executed in partnership with the innovation lab of Cordoba’s government (CorLab).
“Digital transformation is a priority for the government of our city because it allows us to generate key data for decision-making in order to improve citizens’ quality of life. In particular, technologies such as AI allow us to exponentially improve the user experience and the provision of our services. To this end, the link with high-impact local startups has been key, giving speed and quality to this transformation process."
Alejandra Torres, Secretary of Planning, Modernization and International Relations of the Municipality of Córdoba
The city of Cordoba annually collects tens of thousands of citizens’ reviews about their public spaces. These reviews come as unstructured texts and images available online on social media such as Twitter and Google reviews. Taking those contributions into account for decision-making is almost impossible for any human. An AI assistant can simplify this task by understanding, structuring, classifying, clustering and analysing that data and delivering insights to decision makers.
For the purpose of this pilot, the public space chosen is the central area of the city, of approximately 3.5 square kilometres. This is a strategic location to capture feedback since the government is carrying out reorganizations in the public space based on“superblocks.”
The concept of superblocks, piloted in the city of Barcelona, is applied in cities around the world and consists of grouping blocks, pedestrianizing internal roads and expelling automotive transport towards the contours. This seeks to improve the quality of life, making the city compatible with sustainable means of mobility such as bicycles and skateboards, and reducing pollution.
“The selected area for this pilot was attractive to us from a data point of view. Since the points to be surveyed had abundant and relevant information and the implementation of this new urban structure is recent, the platform can solve problems that were not detected through prior channels.”
In 2020, the emergence of COVID-19 resulted in severe impact on citizens’ behavior. The pandemic changed people’s lifestyles, the amount and the type of trips they made, their interaction with open spaces, in addition to the rise of merchants needing to use sidewalks to mitigate a drop in sales.
“I thought it was essential to have a data processing and visualization platform to have access to the real-time sentiment of the population and make decisions based on data. With so much data available today, we need to leverage the new technologies to improve our public services.
In that sense co-creating this project together with UNICEF, OS City and Atix Labs came at the perfect time."
As an example of the relevance of the selected area for the pilot in terms of data, ‘Superblock 1’, comprised of only two squares, the Municipality block and the Court building, had more than 18,500 reviews — 80% of them written during the last two years.
“The objective of this project was to explore the potential of using AI to read and digest the tens of thousands of online reviews about the public spaces in the city, and deliver that information to the government so it can act better, faster, and more accurately.”
Technically, the objectives included:
- Identify the regions to explore (superblocks)
- Identify source types and enumerate them (squares, public buildings, hospitals, etc.)
- Retrieve both georeferenced (Google reviews) and general (Twitter) feedback
- Use A.I. to analyse unstructured feedback from text or geo-tagged images
- Use A.I. to classify, map and visualize unstructured feedback on social media in categories
- Build a web dashboard to visualize the processed information with the potential to share the results with the general public
The pilot started with identifying the superblocks to be analyzed together with the specific spaces to be analyzed (real time pilot here: https://cordoba.socialatlas.app/).
During this pilot, the teams built a web dashboard to view the four selected superblocks with all processed data based on more than 5000 data points. The data points consisted of online reviews from public spaces written by citizens on Google and Twitter.
The data was quickly analysed using A.I. (natural language processing) which helped lower the time government employees would need to manually process this data. More specifically, the A.I. model allowed this analysis 16 times faster than manually.
The outcome of the analysis indicated a very good public reception of these superblocks, taking into account positive appreciations in more than 80% of the categories.
The results also yielded different actionable items, such as the need to increase control and cleaning services over pets and animals in those spaces.
Additional Findings
- Natural language processing without prior training was surprisingly successful in detecting positive / negative comments despite not having the possibility, at this stage, to adapt the algorithm to interpret localisms or phrases with special connotations
- With Google reviews, it was also possible to geo-localize the reviews, enabling the understanding of the relationship between the region's rating and the comments processed.
- The categorization allows grouping results, but in the categories that group few comments, the amount of noise becomes too relevant.
- The original plan, which consisted of first setting the categories of interest and then processing the results, had to be modified since the population does not necessarily generate comments with relevant information for these predefined categories. For example, an original category to detect was Sustainable Mobility, but there were almost no mentions of related elements (means of transport, bicycle lanes, etc). However, it has been very easy to collect information on the state of public spaces such as squares that provide a large amount of data in categories such as sports, leisure, pets, society and entertainment.
Next Steps
After this pilot, the teams are working on a project to implement this technology at the broader city level, while incorporating data from other relevant sources such as tourist information from portals like Tripadvisor and capturing the feedback from Facebook pages of government agencies. The project will also require increasing processing capacity, in-depth training of the model, and integration with the internal incident management system to channel actionable items more quickly.
Atix Labs and OS City are both portfolio companies from UNICEF’s Innovation Fund and pioneers in blockchain and A.I. OS City has great experience working with governments, and Atix Labs, a software development company, has an extensive track record developing and implementing A.I. and data visualization projects. Both companies combine their experience to use open source emerging technology to address citizenship problems.
About UNICEF's CryptoFund
UNICEF’s cryptocurrency denominated investments, the Cryptocurrency Fund, are part of UNICEF’s larger Innovation Fund. UNICEF launched its Cryptocurrency Fund in October 2019 as part of its ongoing work with blockchain technology. UNICEF’s Cryptocurrency Fund and Innovation Fund work together to provide both fiat and cryptocurrency investments to early-stage startups.
