AI, while full of endless possibilities, also presents existential challenges, creating what is commonly referred to as "AI anxiety." 

Just like many of you reading, I feel that anxiety. Sometimes it feels like I’ve stepped into the pages of a science fiction novel, as I look around at our rapidly changing world. From self-driving cars, to impressive but frightening deep-fake photos, , I feel like I’m witnessing technological revolution like never before. Every day, I become more of a broken record as I say to myself (and anyone who will listen): we could, but should we? I ask myself how we can make sure we’re innovating in a way that opens doors rather than replaces jobs

The underlying fear associated with AI is the fear of losing control. However, participating in the session at SSP New Directions, I looked beyond fear, learning new insights on how to embrace change and harness AI as a tool and collaborator.

AI-Generative Text: Are artificial authors the future?

Our recent Research Integrity Survey results showed some hesitancy within our community to fully embrace AI for writing, with 40% saying it should only be used with regulation.

The question of whether machines will become the new authors in academia is definitely a pressing one. Before the session, we were asked to consider some of the most significant pain points related to AI authorship. Three key concerns stood out to me:

Accountability

When a human author publishes a paper, they are accountable for the accuracy of the information presented. With AI, this accountability becomes more complex, as consequences are limited for LLMs.

Data Privacy and Protection

AI systems learn by processing data. To become good at writing scholarly articles, they need to analyze a large amount of content. This raises concerns about data privacy and whether publishers and authors should be informed if their work is used to train AI systems.

Transparency

The process of AI-generated content creation must be transparent. Publishers need tools to detect AI-generated content, and authors should be able to confirm whether AI was used in their work.

The discussion surrounding generative AI authorship is ongoing, and it's essential to address these concerns proactively. With 68% of researchers expressing concerns about the potential for misinformation and inaccuracies resulting from LLMs, there's a clear need for oversight and transparency.

Navigating fear to realize the rewards

Despite fears surrounding this technology, I, and the rest of the Morressier team, believe that AI can play a pivotal role in reshaping scholarly publishing in several ways:

Diversity

AI in authorship has the potential to level the playing field, especially for those who may not be fluent in English. It allows scientists to focus on their work while AI helps them communicate their findings. 

Efficiency

AI can help streamline administrative tasks for authors, like summarizing research or drafting paper sections, saving valuable time.

Collaboration

Imagine an AI continuously updating a taxonomy of all disciplines, helping to identify emerging disciplines and to stay ahead of trends. AI could also be used to predict future research based on past studies, creating powerful opportunities for collaboration.

Conclusion

I think our entire panel could have kept talking about AI authorship, and AI in scholarly publishing, for hours more. If we - as an industry - remain fixed on our goals, we can apply AI in situations that allow us to reap the full rewards of scholarly publishing. 

While, AI-generated text in scholarly articles is a huge topic of debate, it's just the beginning. The possibilities are endless, and the transformation of these fields is an exciting journey.