Why AI Citations Fail: The 58% Problem
A lawyer in Canada faced criminal charges after submitting ChatGPT-generated citations to a judge without verification—and they were fabricated. A Stanford study found even GPT-4 failed to cite federal cases correctly 58-88% of the time. Lavesh Chattlani breaks down why AI can sound confident while being completely wrong, and the one rule that protects you.
From the course “K2's Harnessing ChatGPT” (2 CPE credits, Information Technology, Basic):
https://www.cpetoday.com/course/k2s-harnessing-chatgpt-24c-chatgpt
Taught by Lavesh Chattlani — President at Scalene AI Inc. | AI Strategy, Enablement & Governance, Scalene AI Inc..
More from Lavesh Chattlani: https://www.cpetoday.com/instructor/lavesh-chattlani
CPE Today offers live webinars, Live Replays and self-study for CPAs and accounting professionals. NASBA-sponsored (ID 167619) and an IRS-approved CE provider (ID OWVJT). https://www.cpetoday.com/
#ChatGPT #LegalTech #AIRisks