Reimagining Helsinki: Participatory urban planning with Generative AI

Re-inventing the urban spaces with the help of Generative AI

The Responsible Organisation

The Municipality of Helsinki is the administrative body for Finland's capital and largest city, located on the southern coast by the Gulf of Finland. Helsinki has embraced the Smart City approach with a focus on leveraging technology to enhance urban life while prioritising sustainability, citizen engagement, and innovation.

The city's strategy includes the integration of smart solutions across various sectors such as transportation, energy, and public services. Helsinki is also supporting the creation of collaborative ecosystems where data is openly shared and utilized to improve efficiency and foster new services. Helsinki is committed to achieving carbon neutrality by 2035. The city's holistic and inclusive approach to smart urban development has established Helsinki as a leading example of how technology can be harnessed to create a more liveable and sustainable urban environment for its residents.

UrbanistAI is a new incubated startup that developed a platform through a collaboration between two startups, Toretei and SPIN Unit, bringing together advanced AI development and urban planning expertise.

On the one hand, Toretei is an AI technology company based in Milan and Rome (Italy), founded by an international team of researchers, artists, engineers and designers, and specialised in developing advanced AI (Artificial Intelligence) technologies. On the other hand, SPIN Unit is a Finnish-Estonian private research Lab that creates methods for urban analyses and transformative policies, specialised in urban planning.

By joining forces, they combined their respective strengths, bringing together Toretei's AI skills and SPIN Unit's understanding of how cities are designed and managed. They seek to revolutionise the planning process, enhance efficiency, and develop innovative solutions tailored to current urban challenges by combining research and experience in the urban planning sector with innovations in Generative AI technology. Helsinki 2023 Summer Streets use case is only one of other collaborations with public sector organisations that the UrbanistAI startup has worked on.

The problem

Participatory urban planning is intended to empower communities to shape their environment, but these efforts don't always succeed. The disconnect between public feedback and final decisions, which can stem from bureaucratic obstacles, budget constraints, or differing priorities among stakeholders. Additionally, participation might not be truly representative if certain groups are left out or if there is limited public engagement. When the final urban planning results do not reflect the community's input, it can lead to frustration and a sense of disempowerment among residents, undermining the goal of creating spaces that meet the needs and expectations of all users.

This disconnect was a concern for the City of Helsinki when planning its 2023 Summer Streets project, a major initiative to transform key streets in the city centre by reducing car traffic and adding greenery and street furniture. To make sure the street redesign was a success, the city recognised how crucial it was to consider the views of the people who would be impacted the most: the residents and the local shop owners. The city wanted to learn from them what features and functions these key players desired to be included in the Summer Streets initiative, which generally refers to initiatives where certain streets are temporarily closed to motor traffic to create more space for pedestrians, cyclists, and public activities during the summer months.

The solution and its implementation

The UrbanistAI platform

The UrbanistAI is a cloud-based digital platform that enables public organisations to implement participatory processes by leveraging on the digital twin concept and the potential of generative AI to implement a citizen-driven creative rendering.

A distinctive feature of the UrbanistAI platform is the possibility to train the AI algorithm on local policy requirements. This enhances the ability to co-design urban modifications that align with local policies, such as street furniture placement or traffic and pedestrian regulations. This training is optional because it requires additional budget and the system already includes a certain level of policy compliance validation that works by default, most of the cases without the need for specific training.

Another very distinctive feature is of the platform is the co-design features, that are possible due to its advanced AI-powered drawing and rendering tools. These tools utilise real photos of the targeted urban area and can be used from various devices, including large displays, tablets, and mobile phones. The tools comprise a prompt design web tool, which assists participants in crafting text-to-image prompts and offers pre-defined prompt templates tailored for the co-design process in a user-friendly format. Moreover, an interactive magic drawing tool accessible via touchscreens or pen-based devices can convert rough pen strokes into urban objects (e.g., benches, flower boxes, trees).

Participatory city planning in Helsinki

To address the challenge of ensuring effective participation in the city planning, Helsinki's city planners integrated the UrbanistAI platform into two community engagement workshops held in January 2023. The events were specifically targeted at two key groups: one consisted of a diverse Citizen Committee of 15 members, while the other comprised a collection of business owners from the affected areas.

During the two-hour interactive sessions, participants were divided into small, moderator-led groups. Utilizing UrbanistAI's features, including a prompt design tool and an interactive "magic drawing" tool, each group collaboratively generated multiple design visions for predefined street sections. The platform enabled participants to translate their textual descriptions into realistic images of potential street transformations.

After this co-creation stage, the participants voted on the proposed visions, culminating in a shortlist of the most favoured future scenarios. The workshops concluded with a collaborative dialogue between city planners and participants, utilising UrbanistAI's visualisations to facilitate discussions on the feasibility and potential of the selected designs. This shared understanding, enhanced by the platform's ability to generate realistic representations, proved invaluable in shaping the final design of the Summer Streets initiative. The urban innovation ideas generated during these sessions were later incorporated into the street design, bringing to reality the visions that workshop participants designed with the UrbanistAI platform in the summer of 2023.

The success and potential of UrbanistAI has inspired its adoption by other public administrations seeking to enhance participatory co-design processes. In Smart City Tallin, Estonia, the platform is integrated into the city's participation hub, facilitating co-design sprints between departments and workshops with citizens and relevant stakeholders. Furthermore, UrbanistAI was also implemented for participatory design initiatives in Pristina (Kosovo), cultural projects like the Estonian Museum of Natural History, and urban visioning efforts in Lahti’s City Center (Finland), CityLAB Berlin (Germany), and Zandaam Humankind’s Good Public Space (Nedherlands). This widespread adoption illustrates the tool’s versatility and effectiveness in promoting participation of communities to shape their public spaces across various contexts.

Expected benefits

The integration of AI tools like UrbanistAI into participatory urban design processes promises a series of benefits for both communities and city planners, contributing overall to creating more inclusive and responsive public spaces that reflect the needs of the communities they serve.

By promoting effective participation, the city planning process allows diverse voices to be considered, enabling residents, local businesses, and other relevant stakeholders to actively shape their public spaces. This ultimately leads to a more inclusive city that better adapts to the needs and expectations of all its stakeholders, improving the usability of these shared spaces.
The use of AI-generated visualisations as a collaborative tool fosters an environment where participants can explore and discuss various design options. The shared visual language enabled by the tool helps clarify ideas, enables more concrete discussions, and builds consensus, leading to clearer and more satisfactory outcomes.
Since the AI model is trained on local policies, regulations, and constraints, it allows to identify and address potential technical or legal challenges early on in the design process rather than later. This simplifies the development timeline and ensures that the final voted designs are compliant to local policies.
The participatory use of AI-based tools in this context allows citizens to witness this technology’s capabilities and potential with a real-world application. This can contribute to demystifying AI technology, fostering digital literacy, and building trust in the planning process.

Main challenges

Despite its potential, integrating Generative AI tools into urban co-design presents several key challenges:

User-friendly accessibility: Advanced Generative AI models can be complex. The challenge lies in simplifying the tool, making it intuitive and accessible for users without extensive technical backgrounds. This requires user-friendly interfaces to minimize errors and misuse, eliminating the need for extensive training.
Contextual adaptability: Urban planning and design have unique characteristics and requirements. Generative AI tools must be adaptable to this specific context, ensuring that the generated designs align with local regulations, urban planning principles, and the specific needs of the community.
Democratising AI Use: The goal is to democratise the use of new Generative AI technologies, enabling individuals, regardless of their technical expertise, to employ these tools and benefit from them. The AI should be designed with a user-friendly interface that that empower non-technical users to participate actively in the co-design process.