Quote:
Originally Posted by JABCAT
My university just vetted & launched versions of Chatgpt & Claude for all faculty & staff to use. I've been messing around with it for a little over a week. It's wrong more than it's right on so many things. Then the "I'm sorry, you're right" or the best is when it argues with you, even though it's wrong. It take more time verifying the info it gives you than if you just researched it yourself.
Then you have students using it to write basic papers, which is blantently obvious. And I laughed yesterday when I received an AI-written email from a student (who happens to be a student that has submitted 2 AI-written papers already this Summer term). If it weren't a 6-week class, I'd have referred the student to Dean's Office for academic dishonesty already. I have a feeling that student won't make it very far.
Or, maybe Claude IS a student in my class...  |
It's a real paradox. If you have deep knowledge already, then it's like a team of young analysts doing some basic research for you. I remember that they made mistakes and half of my job was educating them and rerouting them. Claude is the same, EXCEPT that those young users are not getting any guidance and can't tell a bad answer from a good one.
It's most useful when you interrogate, sometimes down several levels. The speed at which it can read through reams of articles and data is incredibly helpful. Also, as a solo practitioner, it's a wonderful reviewer, almost always making useful suggestions. OTOH, when it offers to write something for me and I let it, my rewrite takes more time than just doing it myself.