This week, the great and the good of scholarly communications gathered in Frankfurt for the annual STM Conference. Timed to coincide with the Frankfurt Book Fair, it is the biggest event on the calendar for senior industry members to discuss and debate the big challenges facing publishers, societies, and research institutions worldwide.
The main theme of the day was trust – trust in terms of AI and its uses, trust in terms of research integrity issues, trust in terms of exponential growth in submissions. Indeed, anyone not conversant with the issues of the day in academic publishing would be forgiven for thinking there was something of a crisis of confidence for all those concerned.
Question of trust
Of course, lack of trust in scholarly publishing, or even academia itself, is not a new thing. People tend to forget that peer review is a relatively new phenomenon, being introduced in the middle of the last century in part due to a desire to improve the credibility of academic publications. Before the advent of the internet, there were also numerous publishing scandals that occurred when cross-checking articles for things like dual publication was more difficult.
However, as several speakers at the STM Conference referenced during the day, increasing scepticism of authority, especially online, together with populist voices denouncing academia and ‘experts’, has enabled trust issues to develop. But are these questions of distrust, or of mistrust?
These two terms are often used interchangeably, but there are differences that are instructive for our industry. Distrust is based on clear evidence to the contrary, based on experience or logic with a strong degree of certainty; mistrust is more doubtful, based on ‘vibes’ and tends to be backed by caution or general scepticism about certain things. As such, for many of those in Frankfurt, it felt that there was a mistrust on behalf of delegates when it came to AI and the kind of research being submitted, moving toward distrust as certain data points started to point to specific challenges.
Tipping point?
One of the trusted data points was the sheer number of Chinese-based authors’ articles now being published. It’s been a few years since Chinese output exceeded that of the US, but, more significantly, it also surpassed the US in research funding in the last year. The impact of this growth, however, is still not well understood, particularly the challenges publishers are facing with the recent huge growth in submissions.
Some accredit this to the growth in Chinese scholarship, some to the perverse incentives that have long driven academic outputs and others to the twin integrity issues of paper mills and so-called ‘AI slop’ making it easier than ever to write and submit articles. The truth, as always, is probably a combination of all of the above – with a kind of median position represented by what we might call ‘AI meh’, where Generative AI tools have been used to more easily publish mediocre-but-valid science.
The solution to these challenges to (re)develop trust in scholarly communications was generally held to be through establishing the provenance of research publications and wider collaboration throughout the industry to achieve this. As such, perhaps the most insightful quote of the event came from an economics preprint which has concluded that, “the binding constraint on growth is no longer intelligence. It is human verification bandwidth: the scarce capacity to validate outcomes, audit behavior, and underwrite meaning and responsibility when execution is abundant”. It is gratifying to hear for Cabells which reviews every journal it indexes and lists in its databases with humans, which is still the only model that anyone actually still trusts.
