When scientists add reference numbers to their paper abstract

This is a really bad new habit at least as far as the AI & machine learning literature is concerned. And it is a recent phenomenon regarding AI & machine learning papers.

These reference nos. interrupt the flow of reading and really don't make much sense.

Here is a recent example (not AI or ML related) of such an abstract literally littered/cluttered with reference numbers almost like an obsession:
"The RNA world concept1 is one of the most fundamental pillars of the origin of life theory2,3,4. It predicts that life evolved from increasingly complex self-replicating RNA molecules1,2,4. The question of how this RNA world then advanced to the next stage, in which proteins became the catalysts of life and RNA reduced its function predominantly to information storage, is one of the most mysterious chicken-and-egg conundrums in evolution3,4,5. Here we show that non-canonical RNA bases, which are found today in transfer and ribosomal RNAs6,7, and which are considered to be relics of the RNA world8,9,10,11,12, are able to establish peptide synthesis directly on RNA. The discovered chemistry creates complex peptide-decorated RNA chimeric molecules, which suggests the early existence of an RNA–peptide world13 from which ribosomal peptide synthesis14 may have emerged15,16. The ability to grow peptides on RNA with the help of non-canonical vestige nucleosides offers the possibility of an early co-evolution of covalently connected RNAs and peptides13,17,18, which then could have dissociated at a higher level of sophistication to create the dualistic nucleic acid–protein world that is the hallmark of all life on Earth."