"I decided we would do an oral exam* because it's a great way to see if people have actually learned anything from my course and aren't just parroting notes. Because I can ask them to elaborate on their answers."
Yeah and it's also a great way to get otherwise good students to go blank because it isn't possible to absorb every bit of complex information you spent 12 weeks rushing through, Barbara.
This "gotcha" style teaching fucking pisses me off. There is no time in the real world people are not going to be able to look up their notes. Fuck, half the time I'll ask a professor something and they'll be like "I'll have to look that up later and get back to you." Why? BECAUSE THEY'RE HUMAN AND THATS HOW BRAINS ARE.
This type of teaching only favours students that already had experience with the subject beforehand and freaks with amazing memories. This kind of understanding of the material only comes from experience and repetition, something that the traditional 12 weeks of rushed lectures/labs that discard each topic quickly to fit all of them in don't do.
I fucking hate how much I am going into debt to be taught only the vaguest concepts but doing most of the teaching myself in my own time. Education under capitalism is a joke.
*An oral exam is an exam where instead of answering questions in a quiet room on paper, you have to answer questions on a live video call with your instructor.
Honestly as a self acclaimed AI hater i feel like Animation is one actual decent use case for generative AI. Like you make the art like normal and then have the AI generate extra frames for you that are just slightly edited. Like moving an arm or whatever. But ya just fully generating something is stupid. Hope the animation thing goes well for you whatever you decide to do.
That's already how animation works, except it uses normal math and algorithms to interpolate and generate frames. Frame generation using machine learning is good for generating frames for old film from when really low framerates where the only thing possible.
Pretty much, there's a good video from a year or so ago on why frame generation for animation doesn't work:
https://youtu.be/_KRb_qV9P4g
I found a YouTube link in your comment. Here are links to the same video on alternative frontends that protect your privacy:
No.
I don't think LMMs have a large enough context window to do that well