Aileen Fyfe, in a delightful, fittingly brief history of the abstract [paywalled, alas]:
Many scientists will recognize the sentiment expressed by Royal Society president William H. Bragg in the 1930s: it is vital to have well-written abstracts that can be understood by a “wider circle of readers” (not only experts), yet such abstracts can be “more difficult to write than the paper itself.” Today, such sentiments might generate interest in the possibility of removing this burden from scientists by training large language models to produce summaries of scientific work. In Bragg’s day, the discussion was about whether asking scientists to produce their own abstracts would speed up scientific communication, since it would reduce reliance on labour-intensive abstracting and summarizing services. This is one instance of how the format and function of abstracts have changed over their history, in response to changes in the technologies and practices of scientific publishing.
The author-written abstract took hold only after World War II—like so much else (the good and the bad) we take as ever-thus in scholarly publishing.