« The Secrecy of Certiorari, Trump, and Justice Thomas | Main | 303 Creative, Exclusive Private Enforcement, and Blue-State Revenge »

Monday, March 11, 2024

AI screws up SSRN (Update)

I uploaded a new paper to SSRN, the latest in Rocky and my series on exclusive private enforcement. I would share the SSRN link, but the medical team at SSRN flagged it as having a "medical and/or health term that has initiated a medical screening process;" they removed removed the paper from public view pending further review. According to an SSRN person, the "medical and/or health term" triggering review is "patient"--as in "this law places business owners seeking to exercise a conservative-favored federal right in the same bind that S.B. 8 placed abortion providers and patients seeking to exercise a liberal-favored right."

I assume this is because of an AI filter and not because some human being was stupid enough to make this mistake. But I wonder what would happen if someone wrote a sentence such as "Justice Kagan believes the Court is not patient in seeking constitutional change."

I find it distressing that circulation and publicity of our work is in their hands.

Update: An emailer says this was a simple algorithm rather than "AI." Fair enough and I stand corrected. SSRN still screwed this up. How much? Well, here is the email I got in response to asking to speak to someone with some ability to resolve things:

Your paper or analysis may be framed around a legal, economic or other topic question; however, if the data that is used in the analysis is medical or health related, we must use caution around both patient and health information.
The SSRN system performs an initial automated scan of the submission to identify medical and/or health terms. Depending on the scope of the medical content, SSRN's Medical Team will conduct a deeper medical screening on any such papers that include medical or health data in order to provide complete transparency and to follow best practices around any health data. Due to the caution that is required around health care or medical preprints for prevention of harm and to meet required reporting standards, SSRN screens these papers to ensure they have appropriate declarations around competing interests and funding as well as ethical approval and trial registration, where appropriate.

Your paper is now available on your "My Papers" page and has been expedited.  It is also publicly available on SSRN's website.

Again, we used the word as a noun to describe someone who, seeking a medical procedure, sues to challenge the validity of a law. The word patient(s) appears 7 times, once in the title of a source. Oh, and the paper is not available on SSRN's website or "My Papers;" the link says "This paper has been removed from SSRN at the request of the author, SSRN, or the rights holder." Since I am the co-author and co-rights holder (and I am quite sure Rocky did not remove it), the cause is obvious.

After all this, the article better attract some readers.

Posted by Howard Wasserman on March 11, 2024 at 03:15 PM in Howard Wasserman, Teaching Law | Permalink


The comments to this entry are closed.