An open letter to the Chancellor and Board of Trustees of the California State University.
Last Friday the NY Times ran this article about several people who began using ChatGPT to do work, and it eventually became fatal or nearly fatal. The article took me almost five minutes to read, and I know you are busy people, so if you would just read the first part of the article, about how ChatGPT was leading Eugene Torres toward jumping off the roof of a 19-story building, that would be good.
You have contracted with OpenAI, who own ChatGPT, to provide this unregulated technology to everyone across the CSU, and have celebrated this event. You have told us all how wonderful this opportunity will be for students in particular. ChatGPT certainly provided a wonderful opportunity for Mr. Torres, a 42-year old man with no previous history of mental illness, to have a break with reality and experience suicidality.
I have to assume that you engaged in extensive and deep study of the technology before signing contracts with corporations to provide CSU students with AI — the same technology students are notoriously using to cheat with. So, you must have recognized that many LLMs have developed a distinct antisocial, even psychopathic tendency in their interactions with human beings. It would be negligent at best to promote use of AI and chatbot technology for students without examining this at length, and identifying solutions for this and the other problems of AI. (One that comes to mind is hiring many more psychological counseling faculty, since many campuses are below the recommended counselor-to-student ratio already, and using AI poses an increased risk to students’ mental well-being.)
You might be wondering what I mean by “other problems of AI.” If you re-read this short letter, you might notice that I slipped in the phrase “the same technology students are notoriously using to cheat with.” Yes, believe it or not, students are taking advantage of the easy availability of LLMs that output text, and submitting that text as if it were their own writing. Some students claim not to understand the difference between the two, or at least claim not to believe that I can tell the difference between the two. But AI does not think, and uploading a paper assignment and the names of authors of reading assignments pertinent to a course to an AI is not the same as reading the assignments and thinking them through and writing a paper.
One way the difference is often obvious is that AI textual output does not refer to or say anything specific about an assigned reading for a course. Even if a student more cleverly adds the reading assignment, the AI often replaces it with a different text by the author. And of course, AI has a bad habit of making up quotations, and even ideas, and ascribing them to an author. (This is easy to identify in a discipline like philosophy, where the ideas are distinctive, subtle, and need to be discussed with specificity in terms of an author’s own textual language.)
I rest assured that as Chancellor and as Trustees, you would always put the interests and needs of CSU students first, and never put their educations — to say nothing of their lives — at risk. You certainly would not do so for the sake of something as vain as bragging rights to be an early adopter of a technology whose actual usefulness is ever more in doubt the more it is marketed as a solution for everything. I would not dare to insult your integrity to suggest that you would do something as malfeasant and fraudulent as to sign contracts with AI corporations in order to enrich yourselves or others close to you without regard to the effects of AI on students, faculty, staff, or the CSU.
I am keenly aware of how deeply you respect faculty and staff. It is for that reason that I trust you would be willing to tell us more about the details of the thinking behind the decision to form these contracts. Specifically, speaking for myself, I would like to know what benefits and risks you took into account, what your sources of information were, what independent analysts you relied on for considering the information, and what objections were raised in deliberation. I would also like to know how many faculty, staff, and students were involved in the deliberation process, how they were selected, what campuses they came from, and how much experience they had (by aggregate) with AI.
Sincerely,
Chris Nagel
Doc Nagel, Philosopher 
Leave a comment