Llama Lingo Canvas Information
Infographic:

Interview:
Client / Interviewee Name
Carl Goran


Reasons you chose this topic
Zettelkasten is a mechanism for storing and retrieving thought processes. It uses paper cards, representing thoughts, that are connected using labels, chaining the thoughts together. 

Some domain experts (The domain experts in this case are artificial narrow intelligence products, which replace the paper cards) use these cards to externalize their thought processes, creating, in essence, artificial domain experts. Following the paper cards, inquirers can consult these experts and have their common questions and concerns answered. The more complex a Zettelkasten, or artificial domain expert, is, the more knowledge it contains and the closer it resembles its creator. 
Intel-A-Chat and Llama Lingo create and/or utilize artificial domain experts that are made from artificial neural networks instead of labeled paper cards. Their aim is pedagogical, helping users explore topics through domain experts’ thought processes. While both are interesting from a technical standpoint, both also have clear use cases in any educational setting. For example, schools can use them to help teach students, companies can use them to help train onboarding employees, and hobbyists can use them to introduce themselves to a topic. 


Client Background
Graduated from Wayne State University with a math degree and its first computer science degree.
Worked as a Fortran programmer in manufacturing.
Pursued post-graduate studies in computer science at the University of Michigan-Dearborn.
Worked on R&D for General Motors.
Worked on database systems for Techcenter.
Worked as a corporate data admin for Kelsy Haze.
Managed at least 16 University of Michigan-Dearborn senior design projects.
Judged multiple University of Michigan-Dearborn senior design competitions.
Works as an independent consultant for project management, databases, and neural networks.


Related or existing work on this product
While neither Intel-A-Chat nor Llama Lingo have existing work, both products are related to an application that the client is developing. This application will absorb, or compile and coordinate, a series of artificial narrow intelligence (ANI) products that behave as domain experts for a single topic. For example, one ANI could specialize in baseball cards, another ANI could specialize in baseball history, and the application could absorb both of these ANI, appearing as a domain expert for both baseball cards and baseball history. With enough ANIs, the application would resemble an artificial general intelligence (AGI) product, even though it is not an AGI.


Hardware Requirements
Because Intel-A-Chat and Llama Lingo will be hosted in the cloud using Azure, there are no hardware requirements except development computers that can handle C# development.
 
Challenges you see
Teams will have to learn a lot about the machine learning concepts and technologies that Intel-A-Chat and Llama Lingo will utilize. For example, most students have no experience with Blazor SQL, Azure Cloud, and their integration. Members must be willing to learn and communicate with the team, including the client themselves, for the client is familiar with all these concepts and technologies and is happy to provide some detailed guidance, including samples that the team can learn from. In short, teams must be willing and able to learn a lot of information about complex, foreign topics. 


Benefits you see
Throughout the development process, teams will have great opportunities to learn about popular topics and technologies, such as machine learning and Blazor Azure Cloud, with a knowledgeable client who is familiar with the senior design process. In the end, the teams will have contributed to creating an interesting pedagogical tool with many use cases, benefiting a wide range of people and fields.
Project Details:
Project Name
Llama Lingo


Software Requirements
Llama Lingo is supposed to produce a model and interface similar to the other product, except it is bottom-up, meaning that it starts with extreme analytics and fits them to a topic. For example, one might have complex data about soil acidity in some region. This data will be translated into data structures that ChatGPT 4 or Llama 2 (Meta’s version of chat ChatGPT), our potential artificial domain experts, will analyze. These models will then determine important things that are relevant to their topic from the data. In this case, certain patterns in past soil acidity could indicate something about future soil acidity. Llama Lingo will use, and/or be developed and deployed with, the following significant technologies:
ChatGPT 4 and/or Llama 2
Azure Cloud
Blazor SQL
Visual Studio
C#, HTML, CSS