ChatGPT's Search Struggles: Inaccuracy and Misattribution Persist

TL;DR Summary
Researchers from Columbia's Tow Center for Digital Journalism found that OpenAI's ChatGPT search tool often provides inaccurate news responses, despite its confident tone. In tests involving 200 quotes from 20 publications, ChatGPT gave partially or entirely incorrect answers 153 times, rarely admitting uncertainty. The tool misattributed sources and struggled with accuracy, prompting OpenAI to acknowledge the need for improvements in search results.
- ChatGPT’s search results for news are ‘unpredictable’ and frequently inaccurate The Verge
- How ChatGPT Search (Mis)represents Publisher Content Columbia Journalism Review
- Study of ChatGPT citations makes dismal reading for publishers TechCrunch
- ChatGPT Search Fails Attribution Test, Misquotes News Sources Search Engine Journal
- ChatGPT search provides inaccurate sources even for OpenAI partner publishers THE DECODER
Reading Insights
Total Reads
0
Unique Readers
11
Time Saved
2 min
vs 3 min read
Condensed
85%
404 → 62 words
Want the full story? Read the original article
Read on The Verge