Mission Blackout: AI Decision Authority Simulator

Inspiration

As AI becomes more capable, a bigger question is starting to emerge: not what AI can do, but what AI should be allowed to do.

I was inspired by the challenges of future Moon and Mars missions, where communication delays make it impossible to get real-time guidance from Earth. On Mars, a message can take up to 22 minutes to travel one way. In an emergency, waiting for human approval could mean the difference between solving a problem and losing a mission.

As someone interested in both AI and space exploration, I started thinking about a question that feels increasingly relevant beyond space missions as well: at what point should an AI system be trusted to act on its own?

Mission Blackout was built to explore that question in a simple, interactive way.

What it does

Mission Blackout is an AI Decision Authority Simulator that places users in the role of a crew operating far from Earth.

Users can choose a mission location, trigger emergency scenarios such as oxygen leaks or medical events, and see how communication delays affect decision-making. The simulator evaluates the severity of the situation and determines whether the AI should wait for Mission Control, assist the crew, or take autonomous action.

Rather than focusing on what the AI knows, Mission Blackout focuses on when the AI should be trusted to act.

How I built it

Mission Blackout was built as a lightweight web application using Next.js, React, TypeScript, and Tailwind CSS.

To keep the project focused and achievable within a hackathon, I used a rule-based decision engine instead of relying on external AI APIs. Emergency scenarios update mission telemetry such as oxygen, power, water, and crew health. The decision engine then combines mission risk with communication delay to determine the appropriate level of AI authority.

The result is a fast, responsive simulator that demonstrates how decision-making changes as crews move farther away from Earth.

Challenges I ran into

One of the biggest challenges was avoiding the temptation to build just another AI chatbot.

My original idea centered around an AI copilot for astronauts, but I quickly realized that many AI projects focus on assistants and chat interfaces. I wanted to explore a less-discussed problem: who should have decision-making authority when communication delays make human oversight difficult.

Another challenge was balancing realism with simplicity. Real space missions involve countless variables, but I had limited time to create a working prototype. I had to identify the smallest set of factors that could still communicate the core idea effectively.

I also spent time refining the decision logic so that the outcomes felt intuitive and easy to understand, even for users with no background in AI or aerospace.

Accomplishments that I'm proud of

Take a complex topic involving AI governance, autonomy, and space exploration and turn it into something people can understand within seconds.

The simulator creates a relatively simple implementation. Users can immediately see how the exact same emergency may require different decisions depending on whether the crew is in Earth orbit, on the Moon, or on Mars.

What I learned

Building Mission Blackout reinforced that the hardest part of AI is often not the intelligence itself, but the framework that governs how and when that intelligence should be used.

I learned that autonomy is not a binary choice. In many situations, the best outcome comes from balancing human judgment with AI assistance, and that balance changes depending on risk, uncertainty, and time constraints.

I also learned how powerful storytelling can be when designing a product. Framing the project around a real-world challenge helped communicate technical concepts in a way that feels relatable and engaging.

What's next for Mission Blackout: AI Decision Authority Simulator

The next step is to evolve Mission Blackout from a rule-based simulator into a more dynamic decision-making platform.

I would like to incorporate real AI models, adaptive mission scenarios, and explainable reasoning that changes as conditions evolve. Future versions could simulate longer missions, resource tradeoffs, competing objectives, and ethical dilemmas where there is no perfect answer.

Beyond space exploration, I believe the same framework could be applied to disaster response, healthcare triage, autonomous vehicles, and other environments where communication delays, uncertainty, and high-stakes decisions are common.

Built With

• codex