Research and Generative AI, or; How I learned TL;DR and Stopped Reading Comprehensively
The allure of AI-powered platforms in education is obvious by now. But as of yet, we don’t dwell nearly enough on the wider implications for research.
For instance, SciSpace.io facilitates an unprecedented ease in navigating the vast ocean of academic literature, and Perplexity.ai, harnesses the powers of models like GPT-3.5 and GPT-4, thereby providing accessible summaries of research findings and other types of source material available online. These new capabilities represent a significant leap forward in making academic research more approachable for students and researchers alike. In many ways, AI/semantic search blows Google search and ChatGPT right out of the water.
, English professor at Georgia State University, makes a compelling case for using Perplexity and SciSpace in her class to equip students to take on research earlier in their studies than was possible before. I am confident that students benefit greatly in this particular case.Yet, there is a more fundamental problem that comes with its own set of critical considerations. AI is redefining boundaries across disciplines, and with it more subtly research practices we may take for given. This warrants examination.
Research for the Masses
While the emergence of applications like SciSpace and Perplexity AI heralds shifts in how academic research is approached, promising efficiency and accessibility, they also prompt questions about the balances between breadth and depth of engagement with scholarly work. Just as ChatGPT threw teaching into a complex spiral of interrelated problems regarding our core beliefs and understanding of what teaching really is and should be, so do these apps bring about an unparalleled identity crisis for research.
At the heart of academic research lies the rigorous process of exploration, critical analysis, and synthesis of information. Traditional engagement with scholarly articles involves not just the assimilation of presented facts but scrutiny of the methodology, context, and the argumentative nuances that give research its value. It's a discipline that cultivates critical thinking, analytical skills, and a profound respect for the complexity of knowledge. In fact, its essence is very often the opposite of tidbit snappy communication on X and similar platforms. The abstract is about as reductive as it gets. Until now.
SciSpace and Perplexity AI, by design, aim to streamline the initial phases of research. They offer platforms where vast repositories of academic literature become readily navigable, and complex ideas are distilled into more digestible summaries. Using these platforms raises multiple questions regarding accessibility, copyright and curation (for example, how could you possibly trust the ‘5 Top Articles’ suggested by the algorithm?). They also seem to make the assumption that even the abstracts are too long to grab attention now. ‘Too lazy; didn’t read’ in a research context. Really? And where is the link to the full article?
While SciSpace may aim to democratize access to knowledge, making it more approachable to a broader audience, it clearly also introduces the risk of superficial engagement. The ease with which information can be accessed and understood may inadvertently discourage the meticulous, often laborious, process of deep reading and critical analysis that forms the cornerstone of scholarly work.
Maintaining Depth of Engagement
This concern extends beyond just the individual researcher's engagement with literature. It touches on the broader landscape of knowledge production and dissemination. If the academic community begins to lean heavily on AI for summarization and interpretation, there's a potential shift towards generating content that caters more to AI's capabilities of summarization rather than to the advancement of knowledge.
This scenario may well lead to a dilution of academic rigor, where the emphasis shifts from the depth of research to the breadth of its accessibility. The critical dialogue between researchers, which often leads to deeper understanding and new advancements, may be replaced by an echo of simplified interpretations.
Furthermore, the reliance on AI-generated summaries and interpretations poses a challenge to the development of critical thinking and analytical skills among students. Higher education is not merely about the transfer of knowledge but about equipping learners with the ability to question, analyze, and synthesize information. The traditional model of engaging with full texts—grappling with complex ideas, discerning arguments, and evaluating evidence—is fundamental in cultivating these skills. An over-reliance on AI for navigating academic literature may breed an entire generation of scholars who are adept at interfacing with technology, but less so at engaging deeply with the content it delivers.
Educators and researchers have the opportunity to harness the capabilities of AI in a manner that complements traditional scholarly practices rather than supplanting them. Reading full papers (and a lot of them) absolutely needs to remain a standard in academia, also - and especially - for future generations.
For example, AI-generated summaries could serve as preliminary overviews, prompting further exploration of the full texts. This approach would maintain the efficiency and accessibility benefits of AI while preserving the depth of engagement that is critical to scholarly inquiry.
Concluding Thoughts
The advent of AI applications like SciSpace and Perplexity AI in higher education presents a complex landscape of opportunities and challenges. While these tools offer the promise of making academic research more accessible and efficient, they also raise important questions about the nature of scholarly engagement and the future of knowledge production.
A thoughtful, balanced approach to integrating AI into academic practices is pivotal to help navigate these challenges, ensuring that the potential of AI will enhance, rather than diminish, the quality and depth of academic inquiry.