This guide provides an introduction to AI and Generative AI in the legal field, and has resources discussing how it can be used in law school and in the practice of law.
The technologies involved are changing rapidly. We will try to update this guide on a weekly basis to reflect new developments. If you have any questions please email the law library at loyolalawreference@luc.edu
Before you start any legal research project, it's important to plan and think about important factors such as issue identification, jurisdiction, and key facts. The same holds true when using LLMs such as ChatGPT for legal research: the better the prompt you give the LLM and the better your description of what you want the LLM to produce for you, the better your results will be. The resources listed below provide guidance on how to plan ahead and write more effective prompts when using LLMs for legal research.
You will need to make an account, and you might need to join a waitlist.
OpenAI recently introduced ChatGPT Plus, a subscription service which costs $20/month. It gives general access to ChatGPT without a waitlist, and access to new features.
This Artificial Intelligence Glossary was published on Law.com on March 14, 2023 and provides a "guide to the key terms you need to know and understand when discussing AI."
This post (dated May 21, 2024) from the North Carolina Bar Association's "From the Center" blog provides an overview of six different freely available general purpose Generative AI models, along with a reminder of some of the ethical considerations involved in using these models.
This post (dated April 4, 2023) from the North Carolina Bar Association's "From the Center" blog provides overviews of two ChatGPT competitors: Bing Chat and Google's Bard.
Part of the Law Library's 2024 Thursdays at 3 video series, this video tutorial provides an update on generative AI and the law and a brief overview of Microsoft Copilot, which is available for free to current members of the Loyola community.
ChatGPT's response to the prompt "tell me about ChatGPT in 100 words or fewer", March 1, 2023.
These are some of the issues to think about when choosing how to use ChatGPT in class:
Academic integrity: It is good to be explicit about when and how use of ChatGPT is appropriate in a class, and how students should indicate use of ChatGPT in an assignment.
Privacy: OpenAI trained ChatGPT on information scraped from the internet, without asking consent of the people whose work they used. A link to OpenAI's privacy policy is here. When you interact with ChatGPT, OpenAI collects information including data includes a user's Internet Protocol address, browser type and settings, the date and time of your request, and how users interacted with the site. ChatGPT might also violate data privacy laws. The Congressional Research Service (CRS) has published a report entitled "Generative Artificial Intelligence and Data Privacy: A Primer", which looks at privacy issues and relevant policy considerations for Congress.
Algorithmic bias leads AIs to generate racist and sexist language, even propose discriminatory policies. ChatGPT can generate discriminatory language. This article, from MIT Technology Review, discusses how to see bias in AI image models, and this article, by the same author, addresses proposals to reduce algorithmic bias.
Human Rights: Open AI relied on workers in Kenya to train ChatGPT to filter problematic content. The workers, who were paid $2/hour, needed to review extremely violent, disturbing images and text. Many described the work as "traumatic."
Consent: Requiring students to use ChatGPT would also require that they consent to ChatGPT's privacy policy. Students should not feel pressured.
Confidentiality: People could enter confidential information into ChatGPT. This article from Bloomberg discusses confidentiality alongside other concerns about using ChatGPT in a business setting.
Equity: OpenAI introduced ChatGPT Plus, where subscribers can get faster access and newer features for $20/month. Currently, GPT4 is only available for GPT Plus subscribers.
Energy Consumption: As this article from The Association of Data Scientists notes, massive LLMs require high levels of computational resources and therefore consume high levels of energy. This high energy demand has implications including environmental impact and sustainability.
Copyright - Large language model AIs often pull information from the internet, including copyrighted sources, which could be copyright infringement. Also, how should the product of ChatGPT be copyrighted or attributed? The Copyright Office now has a webpage to post updates on copyright law and policy issues related to AI. Also, the CRS has published a Legal Sidebar on Generative Artificial Intelligence and Copyright Law, which explores questions that courts and the U.S. Copyright Office have begun to confront regarding whether the outputs of generative AI programs are entitled to copyright protection, as well as how training and using these programs might infringe copyrights in other works.
This article from ZDNet (dated June 13, 2024) tests ChatGPT, Microsoft Copilot, and Google Gemini to determine which generative AI tool performs best at different tasks.
In this article from Harvard Business Review (dated December 13, 2024), the authors make a distinction between Generative AI (the use of deep learning neural network models to generate new content) and Analytical AI (AI systems based on statistical machine learning that are designed for specific tasks, such as classification, prediction, or decision-making based on structured data), and then offer advice on when to use each.
