What is generative AI?
At Tufts, generative AI tools are increasingly being used for learning, teaching, research, and administrative work, making it important for Tufts community members to understand what these tools are—and what they are not.
Broadly speaking, generative AI is a computer technology which allows users to generate and analyze text, images, video, code, and other content.
Guidance for Using AI at Tufts
How does it work?
Each generative AI is a computer-based system that can reference a very large set of content known as a data set. This content can be text-based, image-based, or it can be a mix of different media types.
Rather than retrieving or copying specific content, generative AI analyzes patterns in its dataset and uses those patterns to generate new output—such as sentences, paragraphs, images, video, or code.
Some generative AI tools, like Copilot, are all-purpose tools used to analyze and generate many different types of content.
Others are custom‑built AI tools, such as those created using Copilot’s Agent Builder, which can be designed to reference specific information and respond to more focused questions.
Is generative AI the same as a search engine?
Although generative AI can return links in its results, it’s not the same as a search engine. A traditional search engine returns a list of relevant links and descriptions excerpted from those pages, whereas generative AI creates text or other media by finding and reproducing patterns in large collections of data.
Most major search engines are starting to offer a hybrid, combining traditional search with AI to allow for more detail in their results. These include Google with “AI Mode”, DuckDuckGo with “Duck.AI”, and BAIdu with “Wenxin Assistant”.
Even with these hybrid search tools, it’s still important to evaluate sources and verify AI-generated information.
What is generative AI good at?
Generative AI is particularly good at brainstorming, role play, and recombining ideas. Because it’s good at identifying and juxtaposing patterns, it can be helpful for coming up with lists, outlines, and project ideas, and for summarizing existing texts. And most of the major generative AIs can communicate reasonably well in multiple languages.
What can't it do?
Generative AI may not always be good at tasks that require precision and consistency, such as retrieving a specific historical fact, solving a math problem, or returning the same result every time.
Generative AI does not “understand” the content it creates and does not always know whether its generated content is accurate or true. This is why AI-generated content will often need to be fact-checked.
What are AI xhatbots and AI agents?
A chatbot is a generative AI that has been designed to play a certain role and draw from a particular set of content. These can be helpful assistants that are more likely to provide relevant information than an all-purpose AI. An example is the chatbot on a banking website that provides general information to a customer.
An AI agent, by contrast, has abilities beyond reading, summarizing, and generating content. It can interact with a computer environment and execute a series of independent actions on a user’s behalf such as logging into an account, making selections, completing payments, etc.
Generative AI Resources at Tufts
What AI tools can I use?
See our comprehensive generative AI tools list for descriptions and recommendations.
How can I learn to use AI for work?
See our calendar of Tufts trainings, workshops, and webinars to get you up to speed using AI productively at work.
Can students and faculty use AI for coursework?
See Tufts policies and guidelines for information about using generative AI to create course content or do coursework.
Which courses at Tufts focus on AI?
See the Tufts course catalog or a list of courses focused on generative AI.