Generative AI Guide for Law Students: AI Risks

Deepfakes are AI-generated images or videos designed to spread misinformation and deceive, posing significant risks to democratic processes and significant evidentiary issues.  Here you can watch a UC Berkeley digital forensics expert break down political deepfakes:

For a deeper dive and updated resources, see the Deepfakes/Misinformation folder of the GenAI Resources Zotero Guide

GenAI models can produce biased results because these models are trained on large datasets that can reflect existing societal biases. These biases, which can be related to race, gender, sexual orientation, and other protected characteristics, can result in discriminatory or unfair outcomes for clients.

For more information, see GenAI Resources > Risks & Ethical Issues > Bias

Generative AI requires an enormous amount of energy and brings with it serious environmental concerns.  A few statistics:

• One estimate says that the computing power for AI is doubling roughly every 100 days.  
• Each ChatGPT prompt triggers thousands of server calculations, generating heat that must be dissipated using water-based cooling systems.  A 100-word email generated by GPT-4 requires about one bottle of water.  

Resources:

• Power-Hungry AI Is Driving a Surge in Tech Giant Carbon Emissions. Nobody Knows What to Do About It, The Conversation (Jul. 8, 2024)
• Pengfei Li et al., Making AI Less “Thirsty”: Uncovering and Addressing the Secret Water Footprint of AI Models, (2023)
• For additional resources, see GenAI Resources > Risks & Ethical Issues > Environmental Impacts