New Working Paper From AI Disclosures Project: “Real-World Gaps in Al Governance Research: A Safety and Reliability in Everyday Deployments” (Over 9,000 Gen AI Research Papers Reviewed)

From a Summary Post (via Asimov’s Addendum):

We reviewed 9,439 papers in generative AI research covering both the five leading AI corporations and six prominent academic research institutions in the U.S. and found that fewer than one in twenty Corporate AI papers examine what happens after an AI system meets real users and business contexts.

The summary is organized into the following sections:

• Corporate AI labs now set the agenda
• OpenAI’s sycophancy debacle shows the gap
• Why outside scrutiny matters

Direct to Complete Summary + Charts/Tables

Direct to Working Paper
(40 pages; PDF)

Filed under: Journal Articles, News, Patrons and Users