Leader

NST Leader: Of AI and sham science

IF you think "junk" food is bad, wait till you read "junk" scientific papers.

With the advent of Artificial Intelligence (AI), bad actors have moved from food factories to scientific paper mills.

Two per cent of all scientific papers are thought to be produced by AI-supported paper mills, an AFP report quotes a researcher as saying.

As more and more papers flow out of the assembly lines of the paper mills, there will be victims aplenty. But two stand out — academia and science — as people lose trust in them.

Begin with academia. The academic culture of "publish or perish" is partly to blame for making it easy for bad actors to sell authorship to desperate researchers. It was not always like this.

Granted, publishing academic articles in reputable journals is an ancient practice. But it was more of a "publish at your leisure" culture then.

After all, a publication or two were good enough to prove their academic worth.

Not so now, with little to no government funding for such institutions. Universities and research institutions find themselves compelled to rely on the prestige of publications to attract funds.

Malaysian academic institutions, too, have had their fair share of the "publish or perish" culture since 2007, when federal funds started drying up.

No numbers are available on how many AI-assisted academic papers have been published by local academics, but that doesn't mean plagiarised or fake papers aren't a problem here.

A bigger worry is the death of science, which human intelligence helped build, but which machine intelligence is threatening to bury.

Integrity is the core principle of scientific research. Remove this, and there is no basis for us to trust science.

The situation was grave enough that 14 years ago this month, health journalists Ivan Oransky and Adam Marcus decided to set up Retraction Watch, a website dedicated to tracking and publicising withdrawals of scientific literature.

Last year saw 10,000 retractions, wrote Nature magazine.

Noting it to be a sharp increase over past annual retractions, it described the spike as being the tip of the iceberg.

Peer-review fraud was cited as one reason for the slew of sham papers. The less-than-robust academic publishing process must be blamed, too.

Want an egregious example of an introduction that made its way into a scientific journal recently, as quoted by the news agency?

It goes like this: "Certainly, here is a possible introduction for your topic." It has ChatGPT written all over it. How the scandalous sentence got past the scientific journal's editorial team is a mystery.     

Is there a way out of this AI mess? Yes, there is. But first, a point about academic or scientific fakery.

This preceded the Internet age, so let's not pile the blame on AI.

What machine intelligence has done, though, is multiply its impact manifold.

Starting with AI, a good step forward is to develop machine intelligence in integrity-friendly ways. Science without integrity is just a sham. Another is to put an end to the culture of "publish or perish".

There are other ways to measure academic prestige. Excellent teaching is one way to earn prestige. Science needs to be saved, not turned into a sham. 

Most Popular
Related Article
Says Stories