Table of Contents
TerpAI
Service Overview
TerpAI is a powerful tool designed to simplify your interactions with technology and enhance your daily tasks. Acting as your digital assistant, thought partner, and educational resource, TerpAI helps you manage projects, research topics, and get quick answers with ease. It's invaluable for students and educators, providing clarity on complex concepts and assisting with study guides and lesson plans. As a thought partner, TerpAI offers insights that improve decision-making, whether you're brainstorming ideas or analyzing data. Seamlessly integrating into your existing digital environment, TerpAI is secure and reliable, ensuring your data is handled with the highest standards of privacy. With TerpAI, you can work smarter, learn more effectively, and make confident decisions.
Technology Overview
TerpAI is the name for DIT's robust web-based console for managing cloud infrastructure and general-purpose Generative AI (GenAI) Chatbots. The platform and chatbots integrate generative AI capabilities using Generative Pre-trained Transformer (GPT) technology. This platform ensures efficient and secure management of cloud resources and enhances user interaction through AI-driven chatbots. TerpAI Leverages cutting-edge technologies such as Natural Language Processing (NLP), Large Language Models (LLMs), and Retrieval-Augmented Generation (RAG).
Utilizing NLP, the platform enables chatbots to understand and interpret human language with high accuracy, facilitating natural and intuitive interactions. By harnessing the power of LLMs like GPT-4o, TerpAI chatbots generate contextually relevant, coherent responses and provide meaningful insights that enhance user experience and engagement.
TerpAI Chat User Manual
Login
Navigate to terpai.umd.edu and log in with your UMD credentials
Upon first login, users may receive a greeting from TerpAI, followed by the GPT System Guidelines. The Guidelines document helps set the organization’s boundaries and expectations when using TerpAI.
Select a GPT System
Users will have access to different GPT Systems based on their authorization. To determine the current GPT System context, locate the indicator in the top left corner of the main window. The image below shows that the user has selected the “TerpAI Chatbot” GPT System. This is the default chatbot everyone will have access to.
Start Chatting
Start by typing the question within the chat box. Don’t forget to include as much information as possible to receive the best answer. Once the prompt has been entered, click the send icon to submit it to the GPT System. For more information on Prompt Engineering, see the Microsoft Learn documentation1 2.
Responses that are in progress can be stopped by clicking the Stop Generating button.
Switching GPT Systems
Once users start chatting, the selected GPT System is set. Some users may have access to additional specialized chatbots. To switch to a different GPT System, click “New Chat” in the upper left corner of the interface, then select an alternate GPT System. By default, everyone will have access to the “TerpAI Chatbot.”
Chat Without Data
For systems without an associated Data Source (for example, the Default System), the Azure OpenAI model responds with public information from its training data, like ChatGPT.
In this example, a vague question is entered, and the response is quite generic. Because the Default System using the GPT 4 model does not have sufficient information to answer the question, it gives suggestions instead.
Chat With Data
Administrators can create a GPT System that takes data from an internal Data Source, processes it using the Azure OpenAI model, and provides answers grounded in that data.
In this example, UMD calendar documents are uploaded to the data source. This Data Source is then associated with a GPT System, which uses the uploaded data to answer the user’s question based on UMD-specific information. The default system doesn’t have this configuration.
As shown in the example of chatting with an image below, users can also provide text, word, and PDF documents in their conversation. Document size limits could prohibit upload, but limits are increasingly less restrictive and allow larger documents. Restrictions are based on technology capabilities. Those documents are part of the user's conversation, aren’t training material for the bot, and aren’t available to other users like those provided by a system administrator.
The footnote citations can be downloaded by clicking on them if the bot allows it.
Chat with Image
Users can upload or copy and paste an image into a chat with a GPT-4 Turbo with Vision3 or GPT-4o enabled GPT System and ask questions about it. Be sure to follow Prompt Engineering's best practices for images.
Select the paperclip icon and select an image to upload
For example:
GPT-4 Turbo with Vision and GPT-4o can also help users understand a code block.
Chat with Voice
All GPT Systems are capable of Speech to Text (STT) and Text to Speech (TTS).
Speech to Text
Click the microphone icon to start recording.
The icon will turn red to indicate that the application is recording. Once completed, click the microphone icon again to stop recording, and click the send icon to submit the transcribed prompt.
Text to Speech
Once the GPT System has completed its response, click the speaker icon to read it aloud.
Trying saying, “How can I become a good technical writer?”
Key Capabilities of Generative AI Chatbots
Generative Pre-trained Transformer (GPT)
GPT is an advanced AI language model developed by OpenAI that can understand and generate human-like text. It can assist with tasks such as drafting emails, writing reports, and answering questions by predicting and creating relevant text based on the input it receives.
Large Language Models (LLMs)
LLMs, or Large Language Models, are advanced AI systems designed to understand and generate human-like text by processing vast amounts of data. They can be used for various applications, such as customer support, content creation, and language translation, by predicting and generating relevant text based on the given input.
Natural Language Processing (NLP)
NLP allows the chatbot to understand and generate human-like text. It processes and analyzes large amounts of natural language data to respond in a way that is contextually relevant and coherent. This capability enables the chatbot to engage in conversations, answer questions, and perform language-based tasks with a high level of proficiency.
Retrieval-Augmented Generation (RAG)
RAG is a technique that combines the power of large language models like ChatGPT, Claude with external information retrieval. It enables the chatbot to pull in information from a Domain knowledge of data particular to the departments the bot is used in, to supplement its pre-trained knowledge. This is particularly useful for answering questions that require up-to-date or specialized information that is grounded in the curated KB.
Understanding complex queries
AI chatbot is equipped to understand and respond to complex queries. This means the bot can handle multi-part questions, infer the intent behind a query, and provide answers that consider all aspects of the question. This is achieved through sophisticated algorithms that analyze the structure and semantics of the query.
Context retention
This refers to the chatbot's ability to remember and utilize the context of a conversation over up to two exchanges. It can recall earlier parts of the conversation and use this information to make responses more relevant and personalized. Context retention is crucial for maintaining a coherent and logical flow in conversations.
Key Benefits of Generative AI Chatbots
Productivity enhancing use cases
- Answer questions in a conversational manner (with text or voice chat)
- Generate blog posts and other kinds of short- and long-form content
- Edit content for tone, style, and grammar
- Summarize long passages of text
- Translate text to different languages
- Brainstorm ideas
- Create and analyze images
- Answer questions about charts and graphs
- Write code based on design mockups
24/7 availability
AI chatbots provide round-the-clock assistance, answering student, faculty, and staff queries anytime, which is especially beneficial for universities with a large or international student body across different time zones.
Instant response
Chatbots provide immediate answers to user queries, significantly reducing wait times compared to human-operated services. This also reduces the admin staff overload via email and call volume.
Handling high volume of interactions
During peak periods like admissions or exam seasons, chatbots can efficiently handle a high volume of student interactions, reducing the pressure on the staff.
Cost-effectiveness
Chatbots help save on labor costs and resource allocation by automating responses to common queries.
Scalability
Chatbots can easily scale up to handle an increasing number of interactions without needing additional significant resources, unlike human-operated services, which would require more staff.
Key Challenges of Generative AI Chatbots
Understanding context and nuance
AI Chatbots have limited capability to understand the context and the nuances of human conversation. They might misinterpret sarcasm, idioms, or complex sentences, leading to irrelevant responses.
Dependency on quality and diversity of training data
The performance of an AI chatbot heavily depends on the quality and diversity of the data it was trained on. Biases in the training data can lead to biased responses, which can be problematic, especially in sensitive topics.