Attention Archaeologist: An AI That Gets You Off Your Phone

This is my submission for the Hacktoberfest Open-Source AI Challenge — Week 1: Touch Grass.

What I Built

Attention Archaeologist is a local-first AI experience designed to help people step away from their screens and rediscover the world around them.

Instead of asking you to spend more time on an app, it gives you a short observation mission to complete in the real world.

Choose a safe environment and a duration, receive a unique observation challenge, put your phone away, and pay attention to the details around you. When you return, Blind Recall asks you to reconstruct what you noticed without showing you the original mission first.

The idea is simple: technology can do more than capture our attention. It can help us practise paying attention to the world beyond our screens.

How It Works

1. Generate: Get a creative observation mission powered by a locally running language model.
2. Disconnect: Put your phone away and observe your surroundings for a few minutes.
3. Recall: Describe what you remember without seeing the original instructions.
4. Reflect: Reveal the mission afterward and record what surprised you or what you missed.
5. Keep a field journal: Save your observations locally and revisit them later.

Why I Built It

A lot of technology is designed to keep us looking at screens. I wanted to explore the opposite: what if AI helped us spend less time using AI?

Attention Archaeologist turns a language model into a starting point for a real-world experience, rather than another conversation to get lost in.

AI and Privacy

The project uses Python and a locally running Ollama model to generate observation challenges. The goal is to keep the experience private and accessible, without requiring an account, photos, GPS, or recordings.

Generated challenges are checked for obvious safety and quality issues, with fallback challenges available when the local model cannot be used. These checks reduce risk but cannot guarantee that every generated instruction is safe, so users should always use their judgment.

Code

advitCodes / attention-archaeologist

What's Next

I want to improve the variety and quality of the missions, make the recall experience more meaningful, and explore whether repeated observation sessions can help people notice details they would normally overlook.

The bigger idea is to build AI that does not always demand more of our attention, but sometimes gives it back.