The Guardian picked up on this discussion with an article entitled “‘Full-on robot writing’: the artificial intelligence challenge facing universities.” The article provided background links on specific developments in AI writing and describes how universities are responding to new technological developments, noting how some institutions (this article was focused on Australia) are considering such works as plagiarism in their policy statements. It poses the question of how educators should view current developments:

“To put the argument another way, AI raises issues for the education sector that extend beyond whatever immediate measures might be taken to govern student use of such systems. One could, for instance, imagine the technology facilitating a “boring dystopia”, further degrading those aspects of the university already most eroded by corporate imperatives. Higher education has, after all, invested heavily in AI systems for grading, so that, in theory, algorithms might mark the output of other algorithms, in an infinite process in which nothing whatsoever ever gets learned.

But maybe, just maybe, the challenge of AI might encourage something else. Perhaps it might foster a conversation about what education is and, most importantly, what we want it to be. AI might spur us to recognise genuine knowledge, so that, as the university of the future embraces technology, it appreciates anew what makes us human.”

Despite all the hand-wringing, an Inside Higher Education piece written by a professor of a class (“Rhetoric and Algorithms”) outlines the results of an in-class experiment with AI tools, in which the professor encouraged undergraduate students to use as many AI tools as possible to create an essay. The professor found the overall quality of the results to be poor, but perhaps more importantly for an overarching discussion of this topic, students did not like the process using such tools:

“I asked my students to write short reflections on their AI essays’ quality and difficulty. Almost every student reported hating this assignment. They were quick to recognize that their AI-generated essays were substandard, and those used to earning top grades were loath to turn in their results. The students overwhelmingly reported that using AI required far more time than simply writing their essays the old-fashioned way would have. To get a little extra insight on the ‘writing’ process, I also asked students to hand in all the collected outputs from the AI text generation ‘pre-writing.’ The students were regularly producing 5,000 to 10,000 words (sometimes as many as 25,000 words) of outputs in order to cobble together essays that barely met the 1,800-word floor.”

The professor argues that good writers produce better AI output, noting also that can, with such assignments, be effectively used to illustrate to students about the writing submission and feedback process, with the tools providing immediate feedback to students, which motivated students could use to learn. He argues that others worried about plagiarism in their assigned essays can mitigate the risk of AI-generated work by making assignments very specific, and notes also that educators and university policymakers must take developments in this area into account:

I am deeply skeptical that even the best models will ever really allow students to produce writing that far exceeds their current ability. Effective prompt generation and revision are dependent on high-level writing skills. Even as artificial intelligence gets better, I question the extent to which novice writers will be able to direct text generators skillfully enough to produce impressive results.

I would tend to agree with this author, with the current state of technological affairs. I do wonder how current plagiarism tools would be able to track AI-written content, if it’s not in the corpus of comparative texts for a tool, and possible burdens imposed on writing instructors in determining if work is original or not.

And the more I deal with written texts, I feel more than ever that written assignments are crucial to quality higher education. The academic writing process, in my opinion, sharpens students’ skills in many areas, particularly if work is carefully reviewed by instructors with appropriate and constructive feedback. And I agree with the author of the second article, that AI tools can be helpful learning tools (I myself use AI grammar and language tools for this purpose).

I do, however, worry about a world, as alluded to in the first article, in which journalistic content is written by AI. Rather than question of the role of writing in higher education, perhaps we should question where and how AI (not just written output) interacts with the real world, perhaps skewing perceptions.

Graham, S. S. (October 24, 2022). AI-Generated Essays Are Nothing to Worry About. Inside Higher Ed. https://www.insidehighered.com/views/2022/10/24/ai-generated-essays-are-nothing-worry-about-opinion

Sparrow, J. (November 18, 2022). ‘Full-on robot writing’: the artificial intelligence challenge facing universities. The Guardian. https://www.theguardian.com/australia-news/2022/nov/19/full-on-robot-writing-the-artificial-intelligence-challenge-facing-universities

I propose that all journals should, in principle, try their best to have authors and institutions make their raw data open in a public database or on a journal web site upon the publication of the paper, in order to increase the reproducibility of published results and to strengthen public trust in science. Currently, the data sharing policy of Molecular Brain only ‘encourages’ all datasets on which the conclusions of the manuscript rely to be either deposited in publicly available repositories (where available and appropriate) or presented in the main paper or additional supporting files, in machine-readable format (such as spread sheets rather than PDFs) whenever possible. Building on our existing policy, we will require, in principle, deposition of the datasets on which the conclusions of the manuscript rely from 1 March 2020. Such datasets include quantified numerical values used for statistical analyses and graphs, images of tissue staining, and uncropped images of all blot and gel results.

In my own recent experience with journal editing, I can say that having the ability to access underlying data for any study would provide me with more confidence when considering the results discussed in a paper, regardless of discipline.

Miyakawa does recognize the need for more robust data infrastructures for supporting raw data provision by authors and calls for better infrastructures for this:

…institutions, funding agencies, and publishers should cooperate and try to support such a move by establishing data storage infrastructure to enable the securing and sharing of raw data, based on the understanding that ‘no raw data, no science.’

Read more:

Miyakawa, T. No raw data, no science: another possible source of the reproducibility crisis. Mol Brain 13, 24 (2020). https://doi.org/10.1186/s13041-020-0552-2

1. Co-editor for Evidence Summaries at Evidence Based Library and Information Practice, a fully open journal hosted by the University of Alberta (Canada)
2. Engaged in the IFLA Document Delivery and Resource Sharing (DDRS) committee and the organization of the 16th ILDS conference, which takes place every two years. As part of this involvement, program subcommittee members (including myself and NTK’s Sasha Skenderija) created a “light” peer review process for conference papers and decided to publish our conference proceedings openly as an eBook using the existing NTK infrastructure, including a pre-existing repository and creation of an official cataloging record to improve access to the document.

While at times daunting (finding EBLIP peer reviewers during the summer at times seemed like climbing Mount Everest! And who knew some Czech ISBN rules date back to the mid-90s!), both experiences have left me with a clear understanding of what fully moving to an open access environment can mean, when members of various academic communities take it seriously. It all boils down to a community’s rolling up its sleeves and being serious about defining and implementing what scholarly communications/output mean to that particular community plus utilizing (and pushing to expand and improve) existing technological infrastructures.

When this work is done, everyone involved in the process learns: the educational benefits of really engaging in working behind-the-scenes on an open access journal or eBook (in any role: author, publisher, peer reviewer, copy editor, editor) are tremendous, and anyone with an Internet connection interested in the community’s outputs (note: there are still serious Digital Divide hurdles to be solved in this regard) can engage in discussion about the published output.

Stepping back: defining (seriously) what scholarly communications means to a community

I can understand why metrics defining what is considered to be “scholarly output” were developed (funders—including universities and not just grant funding agencies—must have fair and transparent guidelines for academic promotion and distribution of limited grant funding), but I know I’m not alone in feeling that such standards can be misinterpreted, can be overly inflexible, and can be “gamed” by certain players. In the end, each academic community defines its rules.

In terms of contributing to open access movements, it’s up to each academic community to do its soul-searching. Do we want everyone in the world to be able to access our efforts? Why do our journals charge article processing charges (APCs), if they charge them? Are the APCs fair? Will reducing APCs impact quality (or not)? If there’s money involved, where does all the money go?

If it’s hard for your community to answer such questions, it’s maybe time for the community to ask itself why.

Taking advantage of (and pushing to expand and improve) existing technological infrastructures

In thinking about APCs, if a community decides to handle behind-the-scenes work with volunteers (“paid” by learning to be better writers, reviewers, and editors; by gaining academic service activities to put on CVs and to talk about in annual reviews; fame and glory [j/k]), the only “cost” to providing open journal access is the technological infrastructure. EBLIP works with the Open Journal Systems (OJS), a Public Knowledge project, hosted by University of Alberta Libraries; for the ILDS eBook, we used a pre-existing library repository.

Many of these kinds of individual initiatives, at individual institutions or libraries, have sprung up over the past decade. They’re there for members of academic communities to take advantage of, but are often forgotten and/or limited to members of an individual institution. Sometimes, maybe too often, commercial solutions are more user friendly and have broader reach. If we’re serious about open access—and long-term access to information we can trust—maybe it’s time researchers and libraries start asking themselves better questions about solutions which already exist. Why does “Commercial Solution X” work better than our repositories? Are we fast enough to deliver new solutions that can compete with commercial ones? If we’re too slow, why is this the case? How can we do better?

Then there are the even bigger questions: What can we do to change the common issues we face—not just at our institution, but across our global academic community as a whole? How can we share resources to make our efforts accessible to all? How can we serve a greater good better? 

Such questions are easy to lose sight of in the day-to-day sea of academic activities. I’d argue it’s maybe time to stop getting jostled by the waves and to put our hands on the rudder.

Besides simply the number of classes represented here, I think another point this data makes well is how spread across disciplines these classes are. About 50 are taught in computer science departments, but the rest come from information science/studies, communication, law, philosophy, and others.

Fiesler, C. (2018). Tech Ethics Curricula: A Collection of Syllabi. Retrieved from: https://medium.com/@cfiesler/tech-ethics-curricula-a-collection-of-syllabi-3eedfb76be18

Highlighting recent legislation related the US Environmental Protection Agency (EPA), the article concludes by encouraging scientists to remain aware of how regulation potentially influences the research process:

To ignore attempts by politically elected and appointed individuals to dictate how science should be conducted is to betray the very essence of science.

Remote access for NTK patrons

Wagner, W., Fisher, E., & Pascual P. (2018). Whose science? A new era in regulatory “science wars.” Science 09 Nov 2018: 636-639.

Competing interests (both financial and non-financial) are defined as a secondary interest that could directly undermine, or be perceived to undermine, the objectivity, integrity and value of a publication through a potential influence on the judgments and actions of authors with regard to objective data presentation, analysis and interpretation. Non-financial competing interests can include a range of personal and/or professional relationships with organizations and individuals, including membership of governmental, non-governmental, advocacy or lobbying organizations, or serving as an expert witness.

Read more: https://www.nature.com/articles/d41586-018-01420-8

https://beallslist.weebly.com/

Additional Reading

Derek Lowe provides an update, generally, on “predation”:

http://blogs.sciencemag.org/pipeline/archives/2018/04/02/predation

Excerpt:

Barres devoted much of his last decade to publicly describing the challenges he had faced as a woman in science, and offering ways to correct a system that he viewed as fundamentally biased against the advancement of women and minorities. He also called for mentors to be held more accountable for the training and success of their graduate students and postdocs.

Additional Reading:

Barres, B.A. (2017). Stop Blocking Postdocs’ Paths to Success. Nature 548 (7669). 517–519. doi:10.1038/548517a

The central problem is that the entire industry is built to leverage sophisticated technology to aggregate user attention and sell advertising.

https://www.newamerica.org/public-interest-technology/policy-papers/digitaldeceit/