I'll give it a try -- thanks for the tip!

And, yes, I've tried other approaches. OCRmyPDF on Github created a reasonable copy, but even it noted in its workflow that the diacritics were a serious problem.

I'm absolutely open to the Bluesky hive-mind (nest mind?) helping out here. I'm willing to test other platforms. I'd hire students to help -- but my university cut Arabic. So, no students to hire. There are manual fixes to this problem, but they're time consuming (roughly 5-10 minutes per fatwa).

I also laughed when it said these PDFs were a "worse case scenario." There are *billions* of dollars poured into AI platforms. If they are supposed to be good at *anything* it's pattern recognition.

I have almost 1,800 fatwas to process. Probably over 4-6,000 pages. Maybe ChatGPT 5.2 will fix it!

Same! I'd love to see a breakdown.

And, as @jackjenkins.me reports -- we know that Mamdani canvassed religious institutions, observances thoroughly

bsky.app/profile/jack...

In that post, Elsevier promises that humans will continue to be part of the review process. Yes, sure -- so long as universities value that labor. But, increasingly, many will not, and when they are no longer available, I trust that Elsevier will offer further A.I. use as another "optimization."/end

As U.S. universities demand frequent production & citation churn, they will create the conditions that will continue to erode the deliberate, human researcher and subject-centered approach to academic work. Deliberation and time will be replaced by fast-twitch A.I. authored & reviewed articles.

Oh, and A.I. use was "responsibly" disclosed by Elsevier on the article that they used to ... write the article on how AI will be used in future journal article evaluations.

Where does this end up? Having captured the rankings mechanism of U.S. universities, it seems likely that what Elseiver will develop is a new set of products for universities to buy. Prepare to subscribe to Elsevier's A.I. assistant, which will optimize a researcher's submissions for review.

It's quite simple. In an era where universities *don't* value peer review service as part of research, why not have an A.I. "enhance the workflow." Elsevier wants to see A.I. "support" the peer review process.

www.elsevier.com/connect/reth...

Now? With Elsevier buying into the US News and World Report rankings, and convincing the USNWR to more heavily weight faculty "research" as determined by publication and citation "quality," they've secured their investment for the future of academic publishing in an A.I. world. How?

In 1995, Forbes called Elsevier the first victims of the internet in the academic journal era. Twenty years later, Jason Schmitt on Medium noted that Elsevier went from ruin to annual revenues of $25.2 bn, with a higher percentage of profit than Apple. medium.com/@jasonschmit...

Elsevier has spent billions acquiring academic journals and establishing itself as a "protector" and safeguarding academic research. www.elsevier.com/connect/five...