Given the combination of both language and academic skills that I apply to my editing, artificial intelligence cannot provide more than a very limited substitute for my work.  Achieving the same quantity and quality of my editing, would still require, beyond a round of AI, another round human checking and editing.  Furthermore, my covering emails & correspondence and/or discussions with authors cannot be done by a computer at all.  The same applies even more to my presentation training sessions.  They entail fundamentally human interaction in a three-dimensional manner and again, a computer could only replace a small part of this.


A more detailed explanation:

This is an issue which has recently risen to prominence and on which I have substantial related and relevant experience, after many years of high-level work in the field of Academic English.    Many non-native speaking academics seem to believe that they no longer need a real person to edit their work.  Journals too seem to accept this argument.   Such perceptions are certainly very common, but may have more to do with unrealistic expectations and hype than the realities of the situation.

The issue is by no means as simple as frequently perceived, given a variety of factors.  Indeed, ensuring that academic papers are really written clearly and correctly at the appropriate level and using the right style, is a complex and sophisticated challenge.  Certain aspects are beyond the range of AI at present, and it may remain so for the foreseeable future.

Firstly, many articles need both comprehensive linguistic and academic editing.  The latter can be done only partly by a computer, given that such software has clear limitations.  For example, some word choices are extremely subtle and relate not just to the narrow context of the article, but also to the broader usage of a particular word in the field.

I recently had a lively discussion of this short sentence (among many others!) with my students: “People have the possibility to get cheap loans”. Firstly, while not grammatically wrong, the word possibility as used here is “Denglish” – one does write this way in German, but not in English, and get is too colloquial for a formal academic article, as is arguably cheap.  The optimal linguistic and economic solution would be the concise: “People are able to obtain low-cost loans”.  It is extremely unlikely that a computer would come up with this. But nor would an underqualified person or one qualified in the wrong field, and this brings me to the next point.

The standard of academic editing offered out there in the market is uneven, to say the least.  I recently came across very poor samples from two large editing companies.   These editors simply did a kind of linguistic proofreading, entirely overlooking issues which were academically unclear and needed clarification, probably with the authors themselves, as I often do. And some correct phrases which were correct were even made wrong, such as incorrectly inserting “a” before the phrase “fertile ground” (this landed on [a] fertile ground”.   AI could probably match that kind of suboptimal service, and save both time and money, but the result would also be unsatisfactory.

Good academic editing requires all sorts of corrections and improvements that entail not only linguistic skill and subject-related understanding, but also serious thinking and careful consideration. It also often involves multifaceted and interdisciplinary approaches to the text, such as knowledge of both marketing theory and statistical English.  Many changes that a good editor makes are linguistic refinements in the interest of clarity, rather than correcting actual English errors.

Such sophisticated brainwork in the truest sense is simply beyond the capabilities of computers, given that they are still not capable of “productive thought”.  These include the use of commas for academic clarity, the logical order of certain phrases and clauses in sentences, the appropriate phrasing of hypotheses and other vital elements of academic work, such as in questionnaires.  This phrase, “since the beginning of marketing research”, is not really wrong, but “since the early days….” is more elegant. I cannot imagine KI figuring out this kind of subtle improvement.

The difference between “strong” and “weak” AI is also fundamental. Although AI is already very advanced compared to previous years, it still remains “weak” in the sense that it cannot think, it is still just software, even if it learns constantly.  But what does it learn from?  Many published academic articles are not particularly well written for a variety of reasons that I will not consider here.  So the software, just like other academics, often learn poor English or academic errors.  My students often tell me that they saw some clumsy English “in a major book” or top-ranking journal, or heard it from a non-native-speaking professor. The software cannot therefore learn sufficiently and reliably from other work in the area, also because there are so many different contexts, and one cannot count on published work to be correct or consistently well written.

The practical consequence is that texts that are translated by state-of-the-art software, still need proofing, correction and a good knowledge of the source language.  I often use AI to translate emails into German, as it is quicker and easier that writing them immediately in German, but this only works because I can check the result for errors, and they are common. These include blatantly wrong word choices, commas in the wrong places, wrong form of address (too formal), and a completely incorrect style for the context.  Indeed, one reads and hears constantly of poor and incorrect machine translations.

Regarding conference and other presentation training, another activity of mine, the issue does not really arise at all, not yet anyway!   People with good brains, qualifications, and experience will continue to prove their worth.