Personally, I think there’s no way an AO could tell the above essay is chatbot produced. I promise you that’s a better essay than some essays that students have sent to schools, definitely a better first draft than some that I’ve received too.
Something to remember is that essays aren’t all that important in the admissions decision at many schools. People who don’t live admissions everyday also don’t have a good sense of what the range of applications/essays look like. Not all my students listen to what I recommend, which means students have sent unintelligible essays (not an exaggeration) in their apps. And some get accepted to good schools (not talking highly rejectives).
I think these chatbots will be utilized even more for things like term papers at school, and other writing assignments. For example, the poetry that people are posting that’s been created by ChatGBT is amazing. And it gets smarter literally every minute.
I think that most humans are uncomfortable with stuff that is really original. When it comes to creative writing, AI simply provides the ultimate plagiarism - people favorite part of what they have read before.
Ironically, it may be easier to use AI to actually test whether a poem is original, than whether a scientific article is.
On the other hand, if AI is able to write scientific articles, it will be an amazing advancement for science, which will free up a lot of time for the scientists to engage in research. AI cannot really decide what sort of thing interests people, but it likely can be trained to take experimental results and create a comprehensible summary. It can at least help with the Introduction, it can easily create Methods and Results with minimum input, and it can take the researcher’s notes, and create a Discussion and Conclusions segment.
In fact, as somebody who has reviewed scientific articles, AI will probably be able to do a much better job at writing these articles than most people. It will also reduce the workload of reviewers by dealing with the multiple grammatical and stylistic issues that plague scientific papers.
Bottom line, I think that AI can be trained to take over the most boring and often time-consuming parts of science, and free scientists to focus on the parts of science that they like the most.
“As I sat in that lonely McDonald’s on the wrong side of the tracks, staring at that greasy Big Mac, I suddenly felt an irresistible urge to change my trajectory toward two main goals: first, to solve the problem of poverty; and second, to join the fight for animal rights.”
In most of the fields with which I am familiar (non-biomedical biosciences, engineering, etc), faculty do not use these services. The services are generally too expensive, relative to NSF grant amounts, and different NSF directorates often look for different things in a grant, so any grant-writer would have to be familiar with the idiosyncrasies of the different directorates, which often change as the people leading them change (which happens pretty often). For humanities faculty, the amounts are even smaller.
Sometimes the school or university will have a grant-writing support service, but these almost always just edit and revise grants, but do not write them from scratch.
Most academics suck at grant writing, so an AI may actually do a better job than they do.
PS. NIH-supported fields may be different, since their grant proposal requirements tend to be more uniform, and the grant amounts are larger, making the use of grant-writing service more worthwhile. Moreover, in these fields there are many more faculty who are supported by soft money, the the stakes are higher.