Unedited Zoom AI Companion Script
Quick recap
The team discussed the overuse of the term AI and the lack of quality data in the field. They also shared their experiences with using ontologies and a tool called Tinderbox for organizing information. The conversation covered a range of topics including managing personal knowledge and project ideas, documenting complex workflows, and the functionality of adornments in a lab assignment. The team also explored the process of organizing and creating agents in a project, the importance of incremental formalization and the need for documentation, and the structure of their project. Towards the end, they discussed the process of creating agents, the challenges of using the Tinderbox, and the potential use of a visual organizing tool for managing a lecture series.
Summary
AI, Ontologies, and Personal Knowledge Management
David and Michael discussed the overuse of the term AI and the lack of quality data in the field. David shared their involvement in the Entalog Forum and efforts to use ontologies as a solution to various problems. Michael also shared their experience with their thesis, including the learning process, challenges faced, and their use of a tool called Tinderbox to rewrite 200 pages. The potential for Michael to publish a paper on their unique methodology was also discussed. Additionally, Michael led a discussion about managing personal knowledge and project ideas during weekend meetups, and demonstrated a tool called Tinderbox for organizing information. A key question about representing a date in an export template arose, which Michael guided Dave through resolving.
Date Formats, Bloom’s Taxonomy, and Tinderbox Workflow
Michael and Dave discussed various date format standards, with Michael demonstrating how to format dates in different ways. Kim shared that they are working on a project related to Bloom’s Taxonomy, while Win mentioned a complex plan for revising their computer security class using Tinderbox. A discussion about documenting a complex workflow implemented through scattered action code in Tinderbox ensued, with Michael suggesting starting with the end result to better understand the documentation process. Win also presented a top-level outline of plans for a class, which included basic course information, themes, learning goals, teaching techniques, and lab assignments.
Assistance, Workshop, and Tinder Box Confusion
Win was having trouble navigating a document or application, which Michael assisted them with, providing tips on how to use the interface more effectively. Later, Win shared their plans to organize their lecture and lab assignment items thematically, while Michael was preparing to lead a workshop on automating the numbering and labeling of tables and figures in documents. They also touched on the functionality of adornments in a lab assignment and the system’s ability to apply attributes to notes based on their placement in the outline. Towards the end, Frederick expressed confusion about the visual correlation between the graphical and outline views in Tinder Box, which Michael and Win attempted to resolve.
Agent Creation and Management Process
Michael and Win discussed the process of organizing and creating agents in a project. They explored how to manually create aliases and structure them hierarchically. Win clarified that the aliases are not manually created but are action-based. Michael then guided Win on how to create an agent, assign it to topics and lectures, and modify its attributes. They also discussed how to apply a prototype to agents and how to deactivate an agent. Michael suggested that it might be more efficient to use multiple agents for different situations instead of relying on one agent to handle everything.
Project Components Organization and Labeling
Michael and Win focused on the organization and labeling of their project components. Michael proposed using prefixes and attributes to identify different components, which Win agreed to implement. They also discussed the structure of their project, with Michael suggesting a week-based outline with containers for topics, lectures, and labs, which Win confirmed would work for their project. They concluded their discussion with an agreement on the need to clean up prototype attributes and set default values.
Project Development and Documentation Strategies
Michael and Win discussed the development of a project aimed at organizing information and data. They emphasized the importance of incremental formalization and the need for documentation. Win proposed starting the documentation early and keeping it updated, while Michael suggested the use of multiple agents under each week instead of two agents moving everything around. They also discussed the challenges of using the Tinderbox and agreed on the need to separate the map view from other views to avoid confusion. Art suggested naming containers to remind oneself of the view type, with Michael recommending using the map view as the primary means of orienting oneself among notes and ideas.
Data Organization and Visualization Tutorial
Michael gave Win a comprehensive tutorial on data organization and visualization in their software, explaining features such as ‘view use columns’, ‘sort’, and ‘collapse and expand data’. They also explored the use of a visual organizing tool, potentially Tinderbox, for managing a lecture series with multiple dependencies and constraints. Michael emphasized the importance of clear labeling and sorting of data for improved understanding. They also discussed the creation and management of a new attribute called ‘type’ in their project, the process of moving tasks to different weeks, and the benefits of using attribute views for better code management. Michael suggested Win to primarily manage things through the Map View, given the nature of their project. The conversation ended with Michael mentioning the possibility of converting the structure developed in Map View into a hard-coded, nested structure in Outline View.
Agent Creation and Technical Issues
Win and Michael discussed the process of creating agents, with Win preferring to encode it rather than doing it manually. They also addressed a technical issue regarding the defaulting of week numbers to 94 and how to resolve it. Michael demonstrated how to adjust the week number for an agent and proposed hard-coding the week numbers to start from one. They also discussed the need to reset the agent’s edict and the potential for running it once an hour to avoid overhead. They faced a problem where the agent was getting its own weak number instead of its parent’s, indicating overwriting action codes. They decided to work on this issue further. They also discussed adding a new topic in the map view and assigning it a week number.
Note Management and Dependency Direction
Win and Michael discussed the issue of sorting and managing notes in a project, particularly when dealing with a large number of notes. Win raised concerns about the lack of control over note ordering and the potential for complexity. Michael suggested the use of the map view and outline view to manage notes and explained the importance of distinguishing between predecessors and dependencies. They also discussed the potential use of action codes to indicate dependencies between notes. By the end of the discussion, they had agreed on the direction of dependencies in their workflow, with formatting packets being a dependency of actual implementing.
Link Types and Dependencies in Project
Michael proposed the addition of ‘predecessor’ and ‘dependent’ link types to their project and emphasized their significance. They and Win discussed the dependencies between new items one and two and agreed that new item one must occur before new item two. They decided to write action code accordingly, but encountered some technical issues. Win suggested a feature to visually indicate when dependencies in their system are messed up and proposed a constraint for lecture numbers. Michael agreed to consider the constraint later.
Automatic Numbering and Labeling of Research Paper Elements
Michael explained a method for automatically numbering notes and figures in a research paper based on their location in the outline. They demonstrated how to use scripts to assign numbers and how to label figures and tables. They also discussed the use of agents to identify and label all tables and figures in a document. Finally, they showed a text expander that allows for easy cross-referencing of notes. The team agreed that this session was helpful and informative.