Ashutosh Ghildiyal, Vice President of Growth and Strategy, Integra
Better together: why AI tools don’t work on their own
The scholarly publishing ecosystem is rapidly evolving, driven by a surge in manuscript submissions, the emergence of AI-powered tools like ChatGPT, escalating reviewer burnout, and growing challenges to research integrity. AI tools hold promise for addressing these issues by enhancing efficiency and productivity, alerting us to integrity breaches, alleviating stress on stakeholders, and supporting sustainable growth in the industry. But these tools should not be used in isolation; once we understand what AI can and cannot do, it’s clear that for the best outcome, human interaction is a must.
Human Cognition and AI
To fully understand the implications of AI, first we must grasp the nature of human thought, because AI mimics the thinking process. Thought can analyze, synthesize, visualize, articulate, create, and replicate knowledge, but it is not intelligent in itself. The brain functions as an extensive repository of memory, holding imprints of all experiences, cultural conditioning, and acquired knowledge. It is a remarkable recording machine, evolving over millions of years and accumulating collective knowledge. Thinking, as the brain’s dominant psychological activity, is a response generated from this vast storehouse of memory.
Although thought can be highly creative in a certain sense, it remains mechanical, driven by the process of recognition and incapable of understanding the unknown. Its processes involve recognition, classification, verbalization, image-making, recording, and recall. However, thought – and similarly, AI – lacks true perception and functions primarily as a tool for translating, analyzing, and communicating observations. It can also imagine, hallucinate, and fabricate. AI shares these attributes and performs these functions more rapidly and efficiently than the human brain. As AI continues to advance, it is expected to become increasingly adept and faster at processing information, significantly enhancing its capabilities in the coming years.
What AI is lacking, is observation, which is distinct from thinking. It is the first step in the research or learning process. When we observe and pay full attention, we gain insights that are original and new, not dependent on memory. Thought can express these insights, but the insights themselves arise from observation. In the hierarchy of mental processes, observation is primary, and thinking is secondary. Thinking is a mechanical process, while attention is non-mechanical. When we are thinking, we are not truly seeing, and vice versa.
This distinction highlights that, just as thought needs observation to be meaningful, AI requires human direction and observation in order to be ethical, purposeful, and productive. Incorporating this perspective into how we consider, deploy, and form policies around AI-powered tools is crucial.
AI tools that can help publishers address challenges in scholarly publishing
- The Surge in Submissions
The surge in manuscript submissions demands more efficient processing and decision-making. AI plays a pivotal role in managing this increased volume, going beyond simple automation to revolutionize the editorial process.
- Automated Manuscript Screening: AI-driven screening tools are vital for evaluating manuscripts for quality, relevance, and adherence to journal guidelines. These tools help reduce the workload for editorial teams, allowing them to focus on more complex submissions. Broadly, there is a movement towards an editorial process that integrates automation with human oversight. As AI tools evolve, they will become more adaptive, learning from editorial decisions to enhance their screening capabilities continually.
- Reviewer Selection: AI’s ability to match manuscripts with the most suitable reviewers is transformative. By analyzing factors such as expertise, past performance, and current workload, AI can streamline the peer review process. However, ensuring that these algorithms do not perpetuate biases or overlook emerging experts remains a challenge. Future developments should focus on incorporating diverse data points to make reviewer matching more equitable and effective.
- AI-Generated Content
The emergence of AI-generated content adds new layers of complexity to preserving the integrity of scholarly publications, and a number of tools have emerged that are help assess content.
- Plagiarism Detection and Originality Assessment: AI tools are advancing in detecting AI-generated content and ensuring proper attribution. However, distinguishing AI contributions from human-written content is an evolving challenge. As these tools improve, originality assessments may shift from measuring content creation to evaluating ethical research practices.
- Papermill Activity Detection: AI is also making strides in detecting fraudulent activities like papermills. The ongoing battle between fraudsters and detection tools highlights the need for increasingly sophisticated AI. Future innovations may involve behavioral analysis and cross-referencing multiple data sources to stay ahead of those seeking to exploit the system.
- Reviewer Burnout
Reviewer burnout is a significant challenge that AI can help alleviate, but it’s crucial to address this with careful consideration of reviewers’ experiences.
- Automated Review Summarization: AI’s ability to generate concise summaries of peer reviews can reduce the workload for editors and reviewers, but it carries the risk of oversimplification. Ensuring that AI-generated summaries capture the full nuance and intent of the original reviews is crucial for maintaining the integrity of the peer review process.
- AI-Based Translation and Fact-Checking: These tools are vital for ensuring inclusivity and accuracy in scholarly publishing. Their effectiveness depends on the quality of underlying algorithms and databases. As these tools evolve, they are expected to offer more reliable support to non-native English speakers and enhance the accuracy of scholarly content.
AI’s Impact on Key Stakeholders: Publishers, Authors, and Libraries
AI is not just boosting efficiency; it is fundamentally reshaping the roles and responsibilities of key stakeholders in scholarly publishing.
For Publishers: AI is revolutionizing the management of submissions, peer reviews, and editorial decisions. While the current tools are just the beginning, publishers must find the right balance between automation and human expertise. AI offers significant potential for cost savings and efficiency, but it is crucial to maintain the human touch that is vital to academic publishing.
For Authors: AI tools assist authors with manuscript preparation, formatting, and submission readiness. However, as AI becomes more integrated into their work, authors must navigate the ethical considerations of its use. With the line between human and AI contributions increasingly blurred, transparency about AI’s role is essential to maintain trust within the academic community. Authors will rely on guidance both from publishers and libraries around responsible use of AI.
For Libraries: Libraries, as guardians of academic integrity, have a unique role in the AI landscape. As AI tools become more prevalent, libraries must ensure their responsible use. This involves establishing guidelines for AI usage, educating users on its potential and pitfalls, and possibly developing in-house AI tools to support their operations.
Key Takeaways:
- For Publishers: Embrace AI as a tool to enhance efficiency, but remain vigilant about maintaining the quality and integrity of the publishing process. Automation should complement, not replace, human expertise.
- For Authors: Use AI to improve your work, but be transparent about its role. Clearly distinguish between AI assistance and your intellectual contributions.
- For Libraries: Engage in the responsible adoption of AI tools. Advocate for ethical AI usage and support the academic community in navigating this evolving landscape.
Note: The AI tools ChatGPT and Gemini were used to assist the development and editing of this article.
LinkedIn: https://www.linkedin.com/in/ashutoshconsult/
Twitter/X: https://x.com/AshutoshConsult