Recently I experienced what it must be like to be a losing contestant on the television show “The Weakest Link” (even though I have never watched the show, have no idea how it works, and didn’t know it was still on).
I was asked to organize a session at the Potomac and Chesapeake Association for College Admission Counseling conference on ChatGPT and how it might impact the college admissions process. “Jim Jump” and “technological savvy” are generally not found in the same sentence, but I was asked to lead the session because I had been interviewed by Forbes for an article about whether AI will kill the college essay.
The key to a successful conference session is having the right panelists, and I was fortunate to enlist David Mabe, the Director of College Counseling at Woodberry Forest School outside Orange, Virginia (previously in admissions at Davidson and Bowdoin), and Amy Moffatt, the Assistant Vice President for University Admissions at Towson University outside Baltimore. Both did an awesome job of preparing and presenting, and multiple times during the session I found myself thinking that I should go sit in the audience and soak up their wisdom so as to prevent me from opening my mouth and demonstrating my ignorance.
There has been a lot of discussion and speculation about the threats posed by Artificial Intelligence, including ChatGPT. Is this one of those inventions, like the printing press, automobile, and mobile phone, that will change and make our lives easier in revolutionary ways? Is it another example of science fiction becoming reality, or another chapter in the “Fad of the Month Club”? Or, to repeat a question asked in an earlier column, is this the first step in our subjugation by our smarter, and hopefully benevolent, machine overlords?
My initial reaction to ChatGPT focused on risks to the application process, particularly the student essay. How can we trust that an essay was written by the student and not the chatbot, and can we require essays when we can’t verify authorship? I have already seen proposed solutions including handwritten essays and proctored essays. I’m sure David Coleman and his College Board mavens are hoping that one consequence of ChatGPT will be bringing back the SAT essay to increase testing numbers. None of those excite me.
In putting together the session, we decided that we wanted to focus not only on challenges presented by the technology but also the opportunities it might afford.
The first of those opportunities is the ability to save staff time on mundane tasks. Amy experimented and found that ChatGPT can be used effectively to compose boilerplate language for any number of tasks ranging from marketing content to job descriptions and staff surveys to travel planning. One particularly interesting use was having the chatbot generate multiple polite ways to end difficult phone conversations.
ChatGPT is not as good with writing that requires creativity, context, or nuance. There have also been examples of cases where the chatbot confidently and even obstinately asserted as true things that were demonstrably false (this is not an actual example but along the lines of “The War of 1812 took place in 1953”).
I asked it to compose the introduction to our session in the style of this blog. It did so, but failed to capture my wit and sense of humor. Or maybe it captured my lack of wit and feeble sense of humor all too well. I’m sticking with the former.
On the college counseling side, Dave experimented with using ChatGPT to build college lists, including asking it to recommend ten colleges similar to Duke but not as selective. He then asked it to make a table with information for each college, estimate chances of admission of each, and then asked it to plan a hypothetical college trip to several of the schools.
He also found that ChatGPT is particularly good at summarizing things. He sees an opportunity for admissions offices at large universities that receive lots of applications to use the tool to generate a dispassionate summary of recommendation letters in less than 100 words, pulling details, anecdotes, or quotations.
There is also an opportunity to address equity in the admissions process. Counselors in schools with huge counseling loads might be able to generate more substantial recommendation letters than they are currently able to with assistance from ChatGPT (not the same thing as having AI write the letters).
Concern for equity is also one of the challenges. Will access to ChatGPT, particularly as free versions disappear, increase the digital divide, hurting students and counselors without access?
Another challenge for both sides of the desk is change management in our offices. How do we incorporate the new technology into our work, and how do we bridge the divide between generations and between those who embrace ChatGPT and those who fear or loathe it?
ECA is always most concerned with ethical implications, and there are several with ChatGPT. The equity issue raised above leads the list.
ChatGPT makes it easy for students to “cheat” on college essays, missing the point that the essay is not about finished product but about process. The real value of the college essay is as a means for introspection. An antidote for cheating might be programs developed by a Princeton student and by turnitin.com to detect the use of ChatGPT, but those have a success rate of 98-99 percent. Is that enough to risk falsely accusing a student of cheating?
One of the unanticipated consequences of ChatGPT is that the ability to have it summarize and even rate applications may encourage application “gluttony,” the quest to increase applications without increasing staff. In a profession where people are already being stretched thin and where staff burnout is an increasing concern, that would be cruel at best and unethical at worst.
The ultimate ethical issue is one found with any new technology, including genetic engineering. Just because we can do something, does that mean that we should do it? The rapid evolution in ChatGPT and similar technology can very easily outpace the development of ethical and appropriate use guidelines. Just in the past few weeks the White House’s Office of Science and Technology Policy has proposed a Blueprint for an AI Bill of Rights and the European Union has proposed a regulatory framework for the use of Artificial Intelligence. Both are important steps, as there is much we don’t know about both the opportunities and the dangers associated with AI.
Dave Mabe proposed one more opportunity afforded by ChatGPT that I hadn’t considered. Its successful use requires careful and deliberate question-building, and Dave sees that as leading to a new profession, “Prompt Engineer.”
I have been accused multiple times of being really good at posing questions and not as good at providing answers. I now recognize that, like Jimmy Buffett, my occupational hazard has been that my occupation’s just not around. Prompt Engineer might be the career I’ve been waiting for all my life. It certainly beats being a losing contestant on “The Weakest Link.”