S&P Global — FutureChoice.AI

Project showreel

FutureChoice.AI is a data-driven multiplayer game built for S&P Global. It invites players to take on the roles of policymakers, innovators, energy producers and data infrastructure leaders, each representing a different force shaping the future of AI. As a team, they negotiate the trade-offs between progress, governance and sustainability, and the tension between rapid innovation and the power required to support it.

Decisions are made on custom-built encoder dials. With no individual screens, attention stays on the shared simulation: a real-time 3D world that projects every choice forward over twenty-five years. Outcomes range from responsible, sustainable AI development to worlds where infrastructure and energy supply come under strain. Players see how their decisions shaped one another's outcomes and the wider system. At the end of every five-minute session, a personalised printout captures the future they built and the consequences of the path they chose.

FutureChoice.AI debuted at CERAWeek in Houston, one of the largest energy conferences in the world, where it became a standout installation. By translating long-horizon data into a hands-on collaborative experience, it made the pressures behind AI growth tangible to industry leaders and policymakers.

— Valerie Tirella
VP Global Head of Creative,
S&P Global

The rapid rise of generative AI promises to reshape industry and problem solving far sooner than expected. Yet its long-term pressures on energy demand, infrastructure and sustainability remain poorly understood.

FutureChoice.AI was built to make those pressures tangible, turning S&P Global's own data into an interactive experience that positions S&P as a source of authoritative insight. The brief was to help decision makers understand how society might balance AI's opportunities with its risks, and make better-informed decisions about the future.

SetReset began by analysing S&P Global's datasets on AI growth, energy demand, infrastructure capacity and grid load, working alongside S&P's internal experts to ensure the experience reflected real industry perspectives. The research revealed both the pressures within each sector and how those pressures intersect.

Those insights defined the four stakeholder roles: government, AI innovator, data infrastructure provider and utility company. Each role's decisions are bounded by what the actual sector can realistically do, but intuitive enough for a first-time player to navigate.

Branching storylines show how those decisions could evolve over the 25-year simulation, with some paths including weighted randomness to reflect real-life uncertainty. Ongoing research into emerging AI and energy issues, supported by current headlines, helped keep every scenario plausible.

The creative ambition was to transform complex, interwoven data into an experience that felt collaborative, intuitive and visually compelling. This shaped both the storytelling and the interaction model.

A recurring challenge was balancing realism with playability. The cascading effects in the data are genuinely complex and risk overwhelming non-specialists. Narrative logic and simulation weighting were tuned so outcomes felt credible to a sector expert and intuitive to a first-timer in the same session.

To keep friction low, clear narrative staging and animated onboarding help players grasp the mechanics quickly and focus on the decisions and their consequences.

A core design decision was the development of encoder dials as the control method.  Physical controls let players interact without looking away from the shared simulation, so negotiation happens between people, not between people and screens. The result is a game that rewards close attention and collective negotiation.

Representing many branching storylines visually was a key challenge. The world had to be flexible enough to reflect player decisions instantly, so the installation was built in a real-time 3D game engine with assets that update the moment a decision lands.

Set Reset created a detailed asset plan covering all major narrative variations. Every asset (buildings, terrain, infrastructure, vehicles, atmospheric states) was modelled from scratch to a single visual language that holds across radically different futures. Asset development ran alongside simulation design, supported by a flexible pipeline that enabled testing and refinement as the team assessed how decisions reshaped the world.

Custom thermal printers were integrated alongside the dials. At the end of each session, every player walks away with a personalised printout of the future they helped create.

Gameplay

Four players, one big screen, twenty-five years of decisions compressed into a five-minute negotiation.

FutureChoice.AI debuted at CERAWeek in Houston, where it became a standout installation. It ran 509 simulations across the four days, averaging nearly six hours of active use per day.

Player response was strongly positive across the sectors the simulation represented: government, infrastructure, energy, policy. Notable participants included researchers actively working on nuclear fusion, who were visibly eager to win the game. The personalised printouts encouraged ongoing discussion past the booth and served as a meaningful point of engagement alongside the conference programme.

The success of the debut led directly to a follow-on activation at the APPEC conference in Singapore, with regional data swapped into the same simulation framework.

— Nathan Hunt
VP, Head of Content, Digital, Social & Strategic Partnerships
S&P Global