Next step to be super human.
Explore interfaces between AI and human.
This project is experimental/research only at this moment.
Main features::
- Find thoughts from text via basic LLM inference.
- Store different type of data in one PostgreSQL database: table, vector, graph, timeseries.
- Flashcard web portal to interact with stored thoughts:
- Multi-speaker audio play via the the Speak TTS server.
Local deploy:
git clone https://github.com/ittia-research/conscious
cd conscious
# Below are for production, switch `prod` and `dev` if want to start services for development
docker compose --profile utils up -d # Start utility services
docker compose --profile dev down # Optional: prod and dev share some ports and can't start at the same time
docker compose --profile prod up -d # Start in background
docker compose --profile prod logs -f # Follow services logs- Find thoughts (knowledge, ideas) that are not within memory.
- Store new thoughts into AI memory.
- Sync between AI and human memory.
- Memory could belong to a individual, a group, or a organization.
- Memory could be linked together (federation).
- Integrate with existing knowledge bases. For example, Obsidian.
And in doing so, hopefully unlock some secrets about conscious.
- Reading in a lot of times are to find some missing puzzles.
- People might learn better with the Zone of Proximal Development.
- No word salad.
- TPU Research Cloud team at Google
- Google (Gemini, Search)