@qonphuceingey, what are you using chatgpt for in college? This is a genuine question - I think these conversations are very hard to have when we are just talking about generalizations. (And to be clear - while I am definitely anti-LLMs in general, I think the technology can have its uses, and I do think there may be a bit of a moral panic in how some people approach the issue).
That said, if we are talking generalities, I have four main disagreements with your perspective, as expressed above:
The perspective that university is "buying the accreditation that you can do these things."
And I want to start - I understand that many employers do treat a college degree like a purchased accreditation. I also understand that there are (way too many) university administrators that have the same attitude. But I think many of us who have concerns about use of chatgpt at universities also disagree with those people. The thing you are supposed to be purchasing at a university is not the accreditation, but the education itself. The accreditation is awarded as a result of the learning.
Which leads to one of the places where I think specifics would help - what do you mean by "you can do these things." What are the "things" that you should be able to do, from your perspective? Generally speaking, my understanding is that there are two sets of "things" that you are learning in undergrad: (a) critical thinking/reading/writing skills in general and (b) building up a knowledge base, both in general and in your specific focus area/major.
The way I see people generally using/talking about using chatgpt in a college environment is to avoid doing those two thing - that is, to try imitate critical thinking/reading/writing skills in a controlled testing environment and to generate knowledge without actually doing the things to learn.
Which leads to my next two disagreements:
2. You cannot "offloa[d] cognitive tasks" when the cognitive task is critical thinking/reading/writing.
Yes, technology is great at helping us do many things that we used to have to do by hand. And I will go well beyond those examples - I'm a lawyer, and I am so, so glad that I have online searchable databases that come pre-organized and searchable with headnotes and shepardizing tools.
But that doesn't replace actually being able to read and understand a case. Or being able to think creatively in how to apply the arguments in a case to a different scenario. And because chatgpt can't think, it can't do that for you.
This is why, for example, many of my law school exams were open book. We weren't be tested on our ability to know the name of a case. We were being tested on our understanding of the law.
To make this more specific, I can give two examples of how this has affected my work just in the last week. In one case, I needed to make a novel argument. That is, no court had ever ruled on the argument I was making. But because I understood the underlying legal principles, I knew how to find cases ruing on completely unrelated topics - that no algorithm could or would relate - that I could use to make my argument. In another, I watched a brand new attorney try to make an argument in court. She clearly knew what the proper law on point was, and she - in her written filing - had done a good job expressing the relevant facts. But as soon as the judge raised a challenge to her that she did not express, her entire argument completely crumbled. Not because the law and the facts weren't on her side (because they were), but because she lacked the training, experience, and understanding of the law (not just knowledge of the law) to be able to effectively make her argument.
3. You cannot "offloa[d] cognitive tasks" when the cognitive task is gaining a knowledge base.
Yes, we have many tools that so we can offload tasks on machines. But we still need a baseline of knowledge. To give another example of modern tech: I love spellcheck. I rely on it constantly. Y'all have no idea how bad I am at spelling because of how heavily I rely on spellcheck (googling words to make sure I'm using the right one and spelling it correctly - I have done it twice just in this paragraph).
But I'm still glad I was taught spelling in school (even though those were the only tests I ever failed and I'm still mad about it), and glad I didn't have the opportunity to avoid learning spelling by using those tools. Because in order for spellcheck to work, I have to get close enough to the right word for the tech to provide me with the right match. (And again, this isn't hypothetical for me. I usually have to go back and work out a closer spelling of a word multiple times a day because my first attempt is so far off.) And second, because sometimes (frequently) spellcheck is wrong. And I need that baseline understanding of spelling to tell when it is wrong.
And that's for something simple like spelling! When it comes to more complicated topics, you need that basic knowledge base - and the knowledge base you need will change and evolve as tech changes, but you still need to have an understanding of the subject independent of anything that might be generated by the tech.
Which leads to the final disagreement I have with the opinion you express above:
4. Chatgpt can't do most of the things it is advertised to do, or used to do, especially in a college setting. You can't actually "do your job" with chatgpt.
Again, I don't know how you are using it in college. But chatgpt can't think for you. It cannot actually analyze anything. It does not actually know things. It cannot provide you with reliable factual information.
To the extent that companies or individuals are using it to "help" do people's jobs, it is demonstrably doing the jobs worse. Every time I have seen someone claim it helps them do a job, either "the job" was meaningless busy work, they have to use a ton of knowledge and skill and effort (and to be clear, knowledge and skill and effort of the underlying subject) to make it work, or (most often) it doesn't actually do the the job in a remotely acceptable way.
Again, if you are using chatgpt in specific ways in (or outside of) college and find it useful, I would be happy to hear about them and discuss those specifics! But without those specifics, this is why I (and others of "[us] people" are extremely skeptical of the the use of chatgpt and other LLMs in universities.
