Generative AI is a type of artificial intelligence that generates content, such as text, images, video or code in response to prompts.
It utilizes machine-learning, and more specifically deep learning models, to generate new, previously unseen outputs based on patterns in the data it has been trained upon.
The best known models, like ChatGPT, Google Gemini, Microsoft Co-Pilot , are trained on massive datasets, mainly sourced from the internet.
They combine transformer models and, in the case of tools that generate text-based outputs, they leverage Large Language Models (LLMs), trained on massive amounts of text data, which involves learning from patterns and relationships in this data. They are like powerful text prediction engines!
In general, generative AI, works in the three phases of (1) training, (2) tuning, and (3) generation, evaluation and more tuning as outlined in IBM's explanation.
When using this technology for academic work be mindful of policy and guidelines that apply at York University. Read our summary of key guidelines and considerations, including stipulations that apply to using AI-based tools under York's Academic Conduct Policy and Procedures.
There are many different types of tools including text-, image-, code-, and video-generating tools
Some are free, while others are freemium (combine free and paid elements), and some are purely subscription-based.
Many of the well-known generative AI (GenAI) tools in the mainstream are increasingly multi-modal, i.e. they can generate text, images, code and more.
It is very important to critically evaluate outputs for bias or inaccuracies or other limitations. Take care to Verify What you Find and Use by checking out the pointers we provide.
Want to learn what types of tools exist and where to find them? Check out the Types of Tools section of the library’s Using Generative AI to Do Research Guide.
When using these tools for conducting research for your studies (assuming course instructor support for this) a number of benefits and limitations apply.
Above all, before you use them:
Consider whether they are enhancing you learning rather than undermining it.
Critically evaluate the outputs of Generative AI tools for bias or inaccuracies
Check out key Benefits of using Generative AI for Research which include adding efficiency to certain research processes and workflows.
GenAI tools are not without their weaknesses. Build your AI literacy skills by consulting Limitations of using Generative AI for Research which outlines the types of limitations that apply and tips on how you can mitigate against them when using GenAI tools in your work.
Similar to citing books, articles and other resources used in your assignments, it is important to cite any AI-generated content you use.
Check out the Artificial Intelligence (AI) Tools section of our Citing Your Work Guide to get pointers on citing these tools using APA, MLA or Chicago styles.
If you ask GenAI tools to find citations to works on a theme you are interested in, it is always important to verify their accuracy. While citation accuracy is evolving and improving with some tools, it is possible these tools will generate incomplete or fake citations.
Consult the Check Citations for Hallucinations (or Fake Information) section of our Verifying What You Find & Use showcased in our Using Generative AI to Do Research guide.