Digital AI chat bot

impacts

criteria

existing solution

users

pecribed

constriants

self determined

economic

personal

social

login

register

Chat GPT 3.5

python code

Open AI API

SQL database

HTML

SPLC API

Each query will be seen by the app as its own unique conversation that is not linked to any previous questions, parameters or responses

The app will consist of login, register, question and response pages.

Users will select if they want to see the context of the response along with the response, and can select how many k chunks they want for generating the context of the response.

The app will require users to register and login so that the stored conversations can be connected to a user.

python client

user expirence

web aplication

Means that the school will not have to purchase a similar product for students

Takes some of the pressure of the teacher to explain concepts and the students can use the AI

Improves average class grades

Provides an education tool that will help improves students test results

It makes studying for exams quicker and easier

minus

interesting

plus

provieds quick and specific responses to questions

can be used to write many diffrent types of responses, like an essay, science response or a naritive story

has lots of information across the internet

has a tendency to hallucinate when presented with question it finds difucult

prone to spreading missinformation as it draws from incorect or opinionated sources

it has an extensive range of parameters that can be aplied to the query, like tone, audience and structure 

  1. Accept user questions and allow them to specify query parameters like tone, intended audience and length of response

  1. Generate a context using k chunks by submitting the question to a vector database

  1. Build a prompt based on the question, parameters and context

  1. Generate and display a response by submitting the prompt to the OpenAI API

  1. Maintain a record of questions, contexts and responses in a local SQL database

  1. Enable the user the review and rate the quality of the completed conversions

  1. Connect to the vector dataset that contains the Nelson digital solutions textbook

  1. Receive data in one format and transform it into another format to store the data locally

  1. Allow the user the store a rating for the conversations 

  1. Must have a responsive web interface

  1. Be accessible on both desktop and mobile devices through different web browsers

  1. Users must be able to find specific answers to question arising from class work

  1. Users must be able to see the contexts retrieved for each question so they and understand how the retrieval argument generation works

  1. Users must be able to rate the quality of each conversation so they can reflect on the quality of their questions to improve responses over time

input query

view response

view context

set K chunk value

users

Password

Username

userID

conversations

userID

query

Conversation ID

response

context

k chunk vlaue

register page

query page

login page

response page

password

username

password

username

show context

k chunk value

query

response

context

called upon when generating context for a query

hosts the vector database with the Digital solutions textbook

uses the query and context to generate a response

contains the language model

input a query

request context from vector database

register

request response from Open AI API

login

store conversation