AI Writing Tools: A Help or a Threat to AJE Language Editing Services?

Since the first spell checkers of the 1980s, computer users have employed AI writing tools. Over the years, development has kept pace with demand as new AI-based tools and services come online almost daily. 

Not all AI writing tools are the same, though, in purpose or quality. Doing a quick web search for “AI writing tools” returns over 2 billion results, and raises the questions, “Which tool or service do I need for my manuscript,” and “Which one can I trust with my work?”

Like all tools, the positive or negative effects of their use is largely dependent upon how they are created and how they are used.

How do AI writing tools help AJE language editing services?

Here at American Journal Experts (AJE) we combine human expertise from over 400 scientific fields with cutting-edge technologies to provide researchers with the most effective and trustworthy digital services in the industry. We can guarantee that your manuscript will benefit from our combination of human expertise and AI tools because:

1. Our AI is built on professional services expertise. 

It took hundreds of highly-trained editors with advanced degrees to produce the millions of edits necessary to effectively train the AI tools we incorporate into our language editing services. The results of this immense time and effort are apparent in every academic manuscript we edit. 

2. We are science specific.

For nearly 20 years, we have specialized in helping researchers prepare their scientific manuscripts for submission to top journals. Our editors are experts in 447 science-based areas of study and over 2000 field-specific topics. Our AI is a direct product of their professional experience. 

3. Our standalone AI-powered digital editing works to polish and improve your scientific writing.

In less than six minutes, our digital editing service provides 95% English language accuracy to help prepare your manuscript for publication in your target journal. In a Nature Research trial of more than 500 manuscripts, papers using Curie were up to 10% more likely to be accepted than others.

4. Our AI-powered digital editing process serves as a foundational step in our professional editing process.

Once this initial step is completed, our editors will dive deeper into your writing. They can edit more thoroughly, checking for consistency, field-specific terminology, and adherence to journal requirements. This ensures that you receive the highest quality edit possible.

5. We continue to improve our AI models.

By incorporating customer feedback, editing expertise, and field-specific data, our tool is constantly evolving and improving. You will always get the most advanced and up-to-date AI tools in the industry.

How do AI writing tools threaten language editing services?

With the advent of AI-powered text generation, we are witnessing a rise of tools coming to market that can produce writing on scientific topics. This poses a threat to language editing services because:

1. These tools do not create new knowledge, that is, they are not performing research, creating connections, or drawing conclusions.

AI writing tools are based on large language models that produce output by combining existing data to come up with a statistically average opinion. Therefore, they are not capable of offering unique views or identifying new correlations.

Research shows that these AI tools lack the systematic reasoning capabilities necessary for writing a multi-paragraph essay with coherence, selectivity, and originality. This downfall creates major complications when trying to apply the tools to academic and scientific topics. 

2. Likewise, these tools represent the homogeneity in writing; their output reads the same. It lacks originality and the unique tone that only an authors’ voice can produce.

During the writing process, human authors naturally backtrack, edit, and change their work as new ideas and connections come to light. AI tools, however, have a linear thought process that can never produce the same levels of creativity and individuality.

Identifying this lack of humanness is driving current AI content detection methods, like GPTZero. These response tools look for text that is too “perfect”, completely free from errors and lacking the irrationality and quirkiness of human writing.

3. ChatGPT and others generate false citations and do not check for accuracy.

Because AI tools are dependent on finite data sets, they do not have the ability to use outside sources to verify the accuracy of data. For this same reason, they cannot distinguish between credible and non-credible sources. 

In a 2022 preprint study of the AI chatbot, Sparrow, researchers found that up to 20% of its responses contained errors. And, in November of the same year, Meta halted public use of its science-specific large language model, Galactica, just days after it was unveiled, because users had identified so many factual errors in the generated text.

More effects of AI writing tools

1. AI tools have the potential to speed up the publication cycle.

While there are still too many uncertainties and ethical concerns surrounding AI text generation to deem the tool useful in academic and scientific writing, other AI-based tools already have proven track records. 

For authors, literature discovery and mapping tools, like Research Rabbit, and those for content summarization, like Summarizing Tool, can free up time for other research activities. And, using grammar checking tools, like Grammarly, during the writing process, and editing tools, like Curie, after completion, expedite a manuscript towards in-depth editing and journal submission. 

Editors and peer reviewers also have opportunities to access AI tools designed to help manage various tedious tasks. These tools can quickly identify a paper’s subject matter to determine if it falls within the journal’s scope, detect plagiarism and duplicate submissions, and assess the appropriateness of experimental design and statistical analyses. 

2. Journals are scrambling to implement new policies and authorship guidelines.

In response to the recent barrage of AI-based content generation tools, journals are urgently developing guidelines to both stave off the unknown ramifications and to protect the integrity of scientific research.  Though each journal’s policy is unique, they are all based on transparency and accountability as key responsibilities placed firmly upon the shoulders of human authors. 

Nature’s new policy asserts that content created by AI tools, like ChatGPT, is considered a method for conducting research and must be documented in the appropriate section of a manuscript. The tool itself is not an author as it can not be accountable for any portion of the work. JAMA authorship guidelines also prohibit AI co-authorship across its journals and require authors to properly report and cite its use. Authors are expected to take full responsibility for the integrity of any content generated by AI writing tools.