AI Tinkers Project Overview

AI Tinkers Demo: Project Overview

• Project goal: Use a large language model (LLM) to interpret data and create visualizations.
• Aim: Create an AGI (artificial general intelligence) that interacts directly with the human brain.
• Specific goal: Improve data visualization for health data, aiming for a system that gathers data from multiple sources and presents it in a unified, meaningful way.
• Current Challenges: Difficulty in consolidating data from various sources due to scattered screenshots and lack of context in data storage.

Technical Approach

• Project uses an LLM (large language model).
• "Cortona" is mentioned as a fine-tuning model.
• The system will scan data sources to identify patterns.
• Integration of understanding "why" data exists in its current format is planned.
• Specific tools or technologies mentioned:
  • Graph RAG workflow
  • Vector Search
  • Knowledge Graph
  • Generative LLM

Data Sources and Use Cases

• Data sources include:
  • Metadata
  • Google Calendar
  • Building supplies and cash flow
  • Current furniture inventory
  • Google timeline (purchases, etc.)
  • Photos and calendar entries
  • Sleep Health data
  • Google Drive
• Current use case analysis:
  • Understanding daily activities on a timeline.
  • Analysis of activity patterns from calendar data.
  • Pattern recognition in financial data.
  • Identifying trends from screenshots and other data sources.
• Tracking and calendar improvements:
  • Automating calendar updates, e.g., correcting oversleeping entries.
• Personal use case analysis:
  • Dealing with ADHD and dementia prevention.
  • Aims to create a comprehensive data analysis and visualization tool.
• Data visualization techniques:
  • Aims to create visuals such as charts, graphs and image representations.
  • Desired outcome: A comprehensive visualization of personal data linked to intentions and motivations.

Strengths and Weaknesses

• Strengths:
  • Ability to interpret data from diverse sources.
  • Aims to improve data visualization and organization.
  • Potential to automate calendar update tasks.
• Weaknesses:
  • Challenges in consolidating and linking data across different formats.
  • Difficulty maintaining a consistent workflow.
  • Current limitations in processing and drawing conclusions from various data sets.
  • The potential for missing or misinterpreted context in data sources.

User Perspective

• Frustrations with scattered data: Difficulty organizing and analyzing data stored in different locations.
• Desire for clear visual representations: Need for charts and visual maps to understand data better.
• Improved integration of context: Want to understand why data exists.

Personal Interests of the Creator

• Data science
• UI path automation.
• Aerospace/Space Force
• Community relations