AI for researchers

The NC State University Libraries is here to help you utilize Artificial Intelligence (AI) in your research. Learn common definitions of AI terms, view available AI tools at NC State, and email us with questions about how we can help.

Want to talk to a librarian about AI for research? Request a research consultation

Definitions

  • Artificial Intelligence (AI) - a broad field, which refers to the use of technologies to build machines and computers that process data in a way that enables them to make inferences in order to recognize images, respond to spoken or written language, analyze data, make recommendations, and more. (Google Cloud)
  • Generative AI - a subset of artificial intelligence that learns from data to produce new, unique outputs (MIT). Examples of generative AI tools include chatbots such as ChatGPT, and image generators such as DALL-E.
  • Generative Pre-trained Transformers (GPT) - a family of neural network models that uses the transformer architecture, which give applications the ability to create human-like text and content, and answer questions in a conversational manner. GPT models power generative AI tools like ChatGPT, and can be used for a variety of purposes such as creating social media content, building interactive voice assistants, and creating learning materials. (AWS)
  • Large Language Models (LLM) - a category of generative AI model trained on immense amounts of data, making them capable of understanding and generating natural language (IBM). OpenAI's GPT-4, which powers their ChatGPT chatbot, is an example of a Large Language Model.
  • Machine learning - a subset of artificial intelligence that enables systems to learn and improve from experience. (Google Cloud)
  • Natural Language Processing (NLP) - a machine learning technology that gives computers the ability to interpret, manipulate, and comprehend human language. (AWS)
  • Neural network - a type of machine learning process, called deep learning, that uses interconnected nodes or neurons in a layered structure that resembles the human brain. It creates an adaptive system that computers use to learn from their mistakes and improve continuously. (AWS)
  • Central Processing Unit (CPU) - A computer's CPU is the part of the machine that works on a list of things to do and runs (executes) each one in order. In the context of machine learning, CPUs are well suited to smaller tasks that can be carried out in sequence (one after another).
  • Graphical Processing Unit (GPU) - GPUs are the processors on graphics cards, and they are designed to execute a task over many elements of data at the same time. Because they excel at this kind of parallel processing, GPUs are especially useful for this kind of large-scale numerical computation that underlies machine learning approaches.
  • High Performance Computing (HPC) - HPC uses all available computing resources to complete many tasks all at once, in a short period of time. HPC systems are extremely useful tools for machine and deep learning approaches to research questions. They consist of many computing nodes, each equipped with many CPUs and GPUs.

Available AI-Powered tools

There are many AI tools available for both free and paid use. Ithaka I+R provides a regularly updated Generative AI Product Tracker that compares AI-powered tools. NCSU's Office of Information Technology (OIT) keeps a list of Approved Enterprise AI Tools for NC State users.

  • Keenious - an AI-powered tool that analyzes your writing to recommend related research articles from our library databases.
  • Gemini - as of November 2024, students, faculty, and staff have access to Google's Gemini chatbot. Gemini works similarly to ChatGPT. If you log onto Gemini using your NC State account, Gemini will not use or review your data to train its AI models.
  • NotebookLM - helps organize and refine your ideas by analyzing documents and providing summaries, explanations, and suggestions. Login using your NC State account.
  • ChatGPT — ChatGPT is a large language model that uses natural language processing to to converse with users through responding to text-based prompts. It is especially effective for summarizing, brainstorming, drafting, and editing tasks.
  • Microsoft Copilot - Microsoft’s generative AI chatbot is available at no cost to students, faculty and staff by logging in to Microsoft 365 (formerly Office) with your NC State account. To ensure your data are protected, you must log in with your NC State identity.
  • Research Rabbit — Research Rabbit is an AI-powered research tool that helps users discover relevant academic literature. Users add papers, then Research Rabbit uses an algorithm to find similar articles and map their connections. It is built to pair well with Zotero (a citation management tool), and is a free program, but requires a user to create an account to use.
  • Perplexity — Perplexity is an AI-powered search tool that links out to its sources

Copyright for researchers using AI

Copyright impacts AI at several levels: training models, ownership of outputs, and the potential for outputs to be infringing. You need to understand your rights and responsibilities in all of these areas. Learn more about copyright for researchers using AI

AI/ML Developer infrastructure

The vast majority of modern AI models rely on powerful GPUs for their computation and large datasets for training. If you are interested in developing and fine-tuning your own AI/ML models, there are both on- and off-campus options available.

Compute

  • NC State HPC cluster - NC State maintains its own High Performance Computing cluster that is available to all faculty members (graduate students, postdocs and other collaborators can be added to HPC projects owned by faculty). This cluster includes a wide range of GPUs, including high-tier models (H100, A100, L40(s)), that can be used at no additional cost to faculty. If you're interested in learning more about getting started with NC State's HPC cluster to run or train AI/ML models, or if you’d like to explore offloading GPU workloads to the HPC cluster, please feel free to reach out to the Research Facilitation Service (RFS).
  • ACCESS - the U.S. National Science Foundation has established a program to help researchers use some of the most powerful computing systems in the country. Many of these resources are specifically engineered with AI/ML workloads in mind, and include both HPC- and desktop-style interfaces. The RFS can help you learn more about computing resources available through ACCESS and get you started with system credits.
  • Commercial cloud resources - Amazon Web Services, Google Cloud Platform, and Microsoft Azure all offer GPU infrastructure and suites of tools to enable AI/ML workloads. If you'd like to learn more about getting started with commercial cloud resources at NC State, the RFS can connect you with resources and staff to get you started.

Model libraries & training datasets

Note: These are external resources that are not owned or maintained by NC State

  • Hugging Face - Hosts a wide variety of open models for various machine learning tasks, along with code frameworks and open datasets of different sizes.
  • Kaggle - A subsidiary of Google, Kaggle hosts a forum for sharing machine learning models, code and datasets.
  • OpenML - An open library of datasets, algorithms and other machine learning tools to help you get started with your ML project.
  • AWS Open Data registry - Amazon Web Services hosts a collection of openly available datasets from a range of domains and disciplines.
  • Data.gov - Central repository of open data from the United States government. Hosts data, tools, and resources to conduct research, develop web and mobile applications, design data visualizations, and more.
  • CERN Open Data Portal - The European Organization for Nuclear Research (CERN) Open Data portal provides access to over 5 petabytes of particle physics data
  • University of California, Irvine Machine Learning Repository - Long-running collection of databases, domain theories, and data generators that are used by the machine learning community for analysis of machine learning algorithms.

Other tools

  • Ollama: while many commercial LLMs provide user-friendly and programmatic interfaces (APIs) over an internet connection, sometimes funder requirements, university policy, or infrastructure constraints require your research data or model to exist on a local system. Ollama allows you to run LLMs and other AI tools on a research laptop, lab desktop or NC State’s HPC cluster. Contact the RFS for more information.
  • LMStudio - Desktop application for experimenting with local and open-source Large Language Models (LLMs) from repositories like Hugging Face. Working locally with LLMs can help to address some of the data security issues associated with machine learning approaches.
  • NC State's Research Storage - network-attached storage for university researchers, suited for storing the large volumes of data necessary to train AI/ML models. Research storage can be mounted to on-campus machines (or machines connected to the NC State VPN) and accessed via Globus.
  • Globus - transfer and share large datasets between configured endpoints. Compute and Flow options are available for integrating AI/ML models into automated workflows.
  • Containerization - containers are pre-built computing units that include all the necessary dependencies for a service or workflow. Containerizing your AI/ML workflows can facilitate moving seamlessly between different infrastructure. Containerized workflows (including Docker images) can be run on NC State’s HPC cluster through Apptainer.

Risks and limitations

With the rapid increase in availability and usage of AI tools, it is more important than ever to be aware of the risks and limitations of these tools. 

  • Privacy - Make sure to review the data privacy policy for any AI tool you use. Most free AI tools claim ownership over any data you input, including private personal information. This data is mostly used to train and improve future AI models, but is not always limited to that purpose. For example, having a conversation with Google's Gemini chatbot could impact the ads that Google serves you. And of course, like all technology, these tools are not immune to leaks and other types of cyber attacks.
  • Accuracy - Generative AI tools, such as OpenAI's ChatGPT and Google's Gemini, have been demonstrated to output false information. AI researchers refer to these inaccuracies as "hallucinations". For example, a New York attorney recently made national headlines for including nonexistent legal cases in a court filing, which were later revealed to have been "hallucinated" by ChatGPT (MIT). When generative AI tools are used to generate citations or other real-world output, this output must be validated using non-AI sources to ensure accuracy.
  • Bias - AI models have long been demonstrated to include harmful biases in both their training data and outputs. Chatbots and image generators have both been shown to amplify gender and racial stereotypes (MIT). Large language models, such as Open AI's GPT-4, have also been demonstrated to contain various political biases (MIT).
  • Environment - Research has found that AI models can have severe environmental impacts due to the amount of computing power required to train them. The amount of computational power required by AI and ML systems has increased exponentially in the last 5 years (Epoch AI). According to Stanford's 2023 AI Index Report, training OpenAI's GPT-3 consumed enough energy to power the average American home for over a century.

Tutorials and guides

There are many tutorials and guides to using generative AI that are openly available online. Here are some that we've found especially useful. Specific tools may have their own user guides or support services, so look for those if you need assistance with using a specific tool. 

Consultations

Campus AI Groups

Many communities and groups have formed around campus to explore AI-related topics more deeply.

  • Data Science & AI Academy - a Libraries partner, which supports interdisciplinary research, teaching and outreach through data science and AI courses, consulting, research enablement, and programming.
  • AI in Teaching and Learning - a faculty and staff learning community formed by DELTA and the Office of Faculty Excellence to open communication channels and provide collaboration opportunities around generative AI and its uses in digital teaching and learning
  • Center for AI in Society and Ethics (CASE) - an interdisciplinary research hub formed to explore the ethical implications of AI technology, focusing on imminent deployment of autonomous vehicles, large language models and social robots for eldercare.
  • AI with the Friday Institute - a program for educators and educational leaders to leverage AI for deeper learning, support teachers and students in effective uses of AI, ensure equitable access in how AI is incorporated into education, and conduct research on AI integration in instruction.
  • Artificial Intelligence Advisory Group - a closed group dedicated to help NC State lead in the innovation and application of AI.

We're glad! Thanks for the feedback.

Website feedback

What is the issue with this page?

We hear you! Thanks for letting us know.