Complacent Scholars Acquire Another Gadget
It’s already problematic that you relied on AI for your research paper instead of crafting it independently. Now, you can go a step further to obscure the proof of your laziness. A new venture believes scholars require a method to disguise LLM indicators, yet they claim their intention isn’t to promote poor practices among academics. AI enhancers are hardly novel – a quick search online will reveal numerous companies marketing AI that enables AI writing to sound less like it originates from AI. However, none, according to the team behind Academic Humanizer, have been specifically designed for academic papers and funding proposals. Fortunately for the world’s indolent scholars reluctant to genuinely fulfill their responsibilities, that situation has now shifted.
The University of Minnesota’s Perspective
“The issue is that drafts assisted by AI tend to be generic and overly wordy, with … exaggerated wording, and excessively long sentences,” Jie Ding, associate professor at the University of Minnesota and cofounder of MorphMind, the entity behind the tool, detailed in its GitHub readme. “They also stray from the author’s individual voice and sacrifice the exactness that academic work relies on.” An earlier version of this statement, penned prior to a recent readme revision, presents the tool’s purpose directly. Academic Humanizer, the introductory section of the readme stated until an update five days ago, aims to “eliminate the indicators of AI writing from papers and grant proposals, while preserving the precise, evidence-based voice that scholarship necessitates.”
The Functionality of the Tool
The GitHub repository features several examples of text processed through the humanizer for those interested in observing its capabilities. Academic Humanizer is fundamentally a Claude skill, meaning it employs more AI to rectify the issues brought into the academic writing domain by AI. To further enhance its generative outputs’ realism, Academic Humanizer can be directed toward its user’s previous work so that it reflects the “author” of the new paper rather than resembling a bot.
Exercising Caution
As noted by the GitHub repository and Ding, the Humanizer is designed to revise AI-assisted drafts rather than produce new material, so it is not meant to add more AI-generated content onto an already AI-created draft. “Academic Humanizer is a writing-clarity instrument aimed at assisting researchers in articulating their ideas more accurately. It is not intended to create original content or bypass review,” Ding conveyed to GadgetLad via email. The readme also cautions users that Academic Humanizer does not absolve them of their responsibility to disclose that AI contributed to writing a paper or fully authored it.
Academic Deterioration
However, if the tool cannot generate findings, fabricate data, or alter citations, instead merely acting as “an editing aid for clarity and voice,” what implications does that have for weak or mistake-laden content in the original AI-produced draft? It likely means that while the tool may refine the writing, it isn’t designed to validate the foundational arguments or evidence. Subpar content might simply come across as more convincingly human – not an encouraging outlook for a field already overwhelmed by AI-generated clutter. Academics from The University of Surrey cautioned last year that they were being overwhelmed by “formulaic research papers” featuring “superficial and oversimplified … analyses” evidently produced by large language models.
The AI-Contaminated Realm of Research
This predicament has even infiltrated the AI research domain itself, where last month, AI output detector GPTZero reported identifying 100 fabricated references across 51 papers accepted by the Conference on Neural Information Processing Systems (NeurIPS). This does not bode well for academic integrity, already threatened by the escalating use of AI in higher education and academia, nor does it augur well for the quality of college graduates. According to MIT, students who utilize AI for writing essays and papers exhibit lower brain activity than those who abstain from such tools, resulting in diminished retention and less genuine learning.
Conclusion: Tool for the Tech-Savvy
But sure, let’s develop a tool to make AI writing sound more human – that’s undoubtedly the appropriate route to take in this situation.
