Grappling with the topic of AI at Work: CrownShy at the Paris AI Action Summit

In February 2025, CrownShy was invited to support a side event at the Paris AI Action Summit, hosted by France’s Economic, Social and Environmental Council (CESE), a constitutional advisory body that brings together representatives from across civil society to inform government policy and the think tank #Leplusimportant.

The topic for the session was the impact of generative AI in the workplace. The goal was to move beyond conventional conference sessions with one-way speeches and presentations and create an inclusive, interactive and democratic space for discussion.

Designing for deliberation

We brought in our experience with Polis, the open source discussion tool created by the Computational Democracy Project that maps opinions to help find common ground on divisive issues. Working alongside our partners Metagov, AgoraCitizen, and Tournesol, we designed a hybrid process that combined short interventions from 14 expert speakers with live audience participation of about.

CrownShy’s facilitation team drew on experience weaving Polis into large, in-person, agenda-setting events. We set up a live facilitated process that was both authenticated and anonymous, allowing participants to contribute freely while enabling organisers to track perspectives across different sectors: government, industry, and civil society.

During the panel, we periodically added comments to the Polis conversation based on the topics being discussed on the panel. The audience could then vote on these live during the session with their phones, shaping the direction of the discussion. At the end of the session, Polis was left open so participants, both in the room and online, could continue to vote or add their own statements.

This approach enabled us to bridge the gap between the limited seats in the room and the broader online audience and ensured that everyone, regardless of prior AI knowledge, could contribute to the conversation. Participation figures showed that contributions and votes were balanced across both groups, showing that remote participants could be just as engaged as those present in Paris.

Photo credits: Yuting Jiang

Introducing PolisFlow

CrownShy provided the technology for the engagement. We used PolisFlow, a process we’ve been developing to augment the insights and the user experience of Polis.

PolisFlow works as a series of steps:

Knowledge base: participants begin with a short educational module, using videos or background material to build a shared understanding of the topic before engagement. In Paris, the expert speakers provided the foundation of information for deliberation.
Conventional survey: simple survey questions gather demographic or attitudinal data, such as whether someone works in government, business, or civil society. For this we use HeyForm, an open source survey tool, to make this step seamless.
Polis conversation: the survey then automatically transitions participants into Polis, where they can vote agree/neutral/disagree on statements and add their own. Traditional interactive tools like Slido provide snapshots of opinion, but are not well adapted to handling polarisation. Polis, by contrast, is designed to map an opinion landscape in real time, revealing hidden areas of consensus across opinion divides. This made it the perfect choice for a potentially divisive topic like AI and its role in the workplace.

PolisFlow effectively connects the data between survey results and Polis conversations. This innovative approach addresses a common limitation of standard Polis, where using the built-in options for demographic data collection requires framing questions as agree/disagree statements.

At the Paris AI Action Summit, participants began with three short questions (about sector and views on AI regulation) on the HeyForm before joining the live Polis debate. What makes PolisFlow powerful is the way it integrates this survey data with the Polis conversation: when the results were reviewed in the room, organisers could show not only overall convergence and disagreement but also how perspectives varied across sectors, and even observe shifts in opinion over time. This is an example of how CrownShy enhances existing open-source tools, making them interoperable and producing richer, more actionable insights.

What participants said

Over the course of the event, participants engaged with 40 statements distilled from the presentations and discussions, generating 1300 votes on 47 propositions both during and after the summit.

The results revealed areas of convergence and areas of ongoing debate. Broad agreement emerged around several key principles:

AI must be introduced responsibly, with training plans for all workers.
There is a strong need for ambitious European regulation, including conformity testing for AI systems before market entry.
Organisations should establish social dialogue around AI adoption, ensuring decisions are co-constructed rather than imposed.
Transparency, data security, and awareness of long-term risks are essential.

Clear rejection was also visible: participants disagreed with the idea that AI inevitably destroys jobs, that it should replace human management, or that workers should secretly use AI for personal gain.

At the same time, four groups of opinion could be distinguished:

A small “anti-regulation” cluster (5%) focused on rapid adoption.
A strongly pro-regulation group (27%) aligned with European leaders calling for stronger frameworks.
“AI voluntarists” (20%) who were optimistic about AI’s potential for workers.
“Montessori of AI” participants (26%) who emphasised giving professionals the space to adopt AI at their own pace.

Despite these differences, 60% or more converged on the need for democratic debate, ambitious regulation, and shared learning.

Looking ahead

For CrownShy, this collaboration demonstrated how deliberative technologies can scale collective intelligence around high-stakes policy debates. It showed that even on controversial issues like AI, publics are often less polarised than expected, provided they are given the right tools to express their views and find common ground.

We’re proud to have contributed to the AI Action summit alongside our partners. We look forward to continuing to bring structured, participatory methods to communities, in-person events and global conversations on technology and democracy.

Are you thinking about how your group could access its collective intelligence?

There is a world of possibility for increasing the accessibility and impact of your group’s interactions. Hybrid events that blend physical and digital participation can connect remote stakeholders who can't attend in-person and gather asynchronous insights from diverse voices across different time zones. Whether you're planning an organizational strategy session, a tricky community <> government consultation, or a large-scale public engagement, our tools can help you access collective intelligence in ways that go beyond what traditional meetings offer.

If you are interested in finding out more about the tools CrownShy is developing or you’d like to learn more about PolisFlow feel free to get in touch.