SPINLAB @ MSU

Nature Reviews Bioengineering published an editorial last year and a comment recently about the use of generative AI for the writing process. In my view, both articles are broadly in agreement about the fundamentals:

Writing compels us to think — not in the chaotic, non-linear way our minds typically wander, but in a structured, intentional manner. [Editorial]

[A]sking a model to write a paper based on minimal prompting, with no further engagement, is rightly considered an indefensible form of cognitive ‘offloading’. [Comment]

All that is not to say LLMs cannot serve as valuable tools in scientific writing. For example, LLMs can aid in improving readability and grammar, which might be particularly useful to those for which English is not their first language. [Editorial]

The Comment by Earp, Guernon, & Mann makes some very good points about the appropriate use and added benefits of using generative AI to aid the writing process:

1. “stress-test claims by soliciting counter-arguments across multiple AI systems”
2. “use AI-assisted literature searches to assess originality by reading flagged sources”
3. a critical evaluation of the text output from an AI model requires demanding intellectual engagement
4. AI models can produce outputs that can be thought of as prototyping to quickly rule out bad ideas or incomplete arguments
5. “researchers with…motor or vision impairments, repetitive strain injuries, dyslexia or attention deficit disorder” can use “AI-assisted drafting…as an accessibility tool that enables their thinking to reach the page”

That said, I stand by my claim that generative AI for an unskilled and untrained user is the equivalent to a loaded self-firing gun handed to a baby.

Tools, in general, are only useful to the skilled user. No tool, enabled by no technology, exists without some caveat or limitation that the user should bear in mind. [Kyle Godbey and Pablo Giuliani]