Will ChatGBT change College. If so, how? Will it make College obsolete in some fields?
There are a lot of implications and a lot of unknowns.
Will ChatGBT change College. If so, how? Will it make College obsolete in some fields?
There are a lot of implications and a lot of unknowns.
When calculators took over doing math for students, I recall similar worries. I think every technological advance brings concerns, but we humans are fortunate to have brains that can help us figure out how to deal with them. I am going to put my faith in our ability to figure out how to coexist with AI.
This is different. Calculators, ultimately, have allowed students to advance further in math by performing more basic functions. In other words, the calculator enhances mathematical skills rather than simply performing them. But Chat GPT is different because it does the writing for you. Now, it’s not good writing – at best, it’s like a decently written Wikipedia entry, hitting the main points but lacking depth or analysis, and at worst, it’s like a champion BS job. It all depends on how the question is written. But in fields that depend on writing, it’s the writing itself that’s evidence of thinking. If we’ve got an app that writes for students, then they’re not researching, they’re not reading, they’re not integrating evidence and context, and they’re not writing for themselves – they’re not doing the necessary thinking and learning that the discipline requires. At best, I can see this maybe as a study aid for students who are trying to learn basic concepts and terms (but then, why not do the assigned readings, which are probably more challenging but also more enriching?). At worst, it’s a shortcut around the analytical thoughtwork that allows students to learn.
Here is my question…is this AI smart enough NOT to generate exactly the same essay (for example) more than once? What if two people put in the same info…
Seems unlikely that it would generate the same response twice, even with the same question, because it draws from an enormous range of existing web content. Someone (a Princeton student, I think) has developed an app that detects ChatGPT responses based on the absence of typical traits of original writing.
It does not generate the same answer twice, although there can be similarities in the output.
ChatGPT does not have access to the web (so it won’t have current events in its repertoire, for example), it only can work with the information the OpenAI team taught it. OpenAI provided a ChatGPT detector well before the Princeton student created his app. Neither detector is 100% accurate.
IME ChatGPT is generating some very good college essays, better than many, maybe most of my students, especially if one considers it a first draft. The reality is that many people don’t have good writing skills. We have already seen the UCAS application (UK schools) drop the personal statement requirement for next year’s college applications, perhaps we will see some US colleges who care about the essay (most don’t) drop their requirement too.
Big picture ChatGPT and other AI tools are going to change education at all levels because of its vast number of uses.
This could be the tool that students use as a first run…but they still need to read and personalize essays more would think.
How about what most other countries do and we used to as well? Make the kids sit and take exams in person with a proctor or two?
That was called the SAT/ ACT. Apparently many wish to not use it. Since grades are all inflated,admissions essay will be written by AI, and SATs not taken, college admissions will be more of a lottery than ever.
I will continue to assign papers because I believe they are essential. I teach history, and the process of researching and writing is more important than factual recall. I have always designed paper topics that are difficult to plagiarize, or for which you can’t buy papers online – these topics are also not as effective at generating ChatGPT responses. In the post-quarantine semesters, in-person exams have gone out of style, because everyone got accustomed to take-home exams in remote learning, but I resumed in-person exams when we went back to campus, in classes for which such exams are appropriate. Seemed like I was among the few who did that, but I suspect I’ll have more company now.
Well, I could well be in the minority here but the way I see it, your average sophomore Hist 102 kid should be able to write a decent essay on John Calhoun and the Missouri compromise based on learned material. Research skills are very important but IMO, those could well be reserved for upper division classes and capstone/thesis projects. They should be followed up by an oral defense of the thesis.
Few schools require a thesis to graduate, even fewer require an oral defense of that thesis. For the vast majority of undergrads at the vast majority of colleges AI will equal or exceed their own work.
Well, the average sophomore cannot necessarily write that essay, but I’m not assigning it, either. The essays I assign in intro classes are designed to engage primary sources in depth to build interpretive and writing skills. When I say “research,” I’m using the term loosely to mean working with primary and secondary source material to develop an argument. I find that this is challenge enough for most of my students (though of course the intro students get more guidance and structure, and the upper-level students have more latitude). But if they’re not doing that work on their own, however flawed it might be, they’re not learning the essential skills they should be getting out of a history class. Not until senior seminar do my students tend to do the kind of original research expected for a capstone – but if they’ve been relying on AI to write their papers up to that point, there’s no way they can manage the capstone.
If the AI has access to the same information as your capstone students, it will soon be able to out-research the student (and may be able to do so now).
Over a decade ago Marc Andreessen said “software is eating the world”. Only a few years later, Jensen Huang said “AI is eating software”. What’s next?
Return to proctored testing is the simplest answer but we as a society do not seem much value in it anymore.
This would worry me as an educator.
Absolutely true, but the goal is to measure success in an unbiased fashion.
Exactly my point. Assign readings and learning concepts for FR/SO students and test them on learning the material and being able to construct a cohesive answer. Leave the more nuanced and difficult research topics for the later years.
Hence, make them take a test.
You seem to misunderstand what is involved in teaching history. It’s not really the point in this thread, but simple factual recall is not it. Building interpretive skills is. You can find facts on your phone, but that’s not the core of history as a discipline.
For now, AI builds responses based on the information you give it. Capstone students are developing individual research topics using highly tailored sets of evidence. Maybe one day, there will be some kind of massive search engine that can find historical sources (I’m not talking something like WorldCat), but for now, all of the archival and digital databases are organized in such idiosyncratic fashions (and require individual approaches to keyword searches, etc.) that the existing AI couldn’t write a properly sourced analytical paper. Thank goodness.
I was an engineering major, but American History was my favorite class in college. The prof was amazing. He assigned us SO much excellent reading. I still remember the final exam question. I probably don’t have it exactly right, but we were to compare and contrast Reconstruction with the building of the Panama Canal. The professor said I should switch majors to history, but I knew I wasn’t cut out to be a teacher so I didn’t.
You can go into a lot of careers with a history major besides teaching, but presumably you found the path that was right for you.