I was a teacher assistant circa 2023.
I required every student to submit the assignments typed digitally, since grading a single handwritten one would take as much time as 5 typed.
I saw a single fully AI generated assignment and it was laughable and an easy zero.
I wonder how I would do it now.
Lot's of people say that's a mess to maintain and too broken to actively use.
I often doubt if that's due to actual problems, or mix with that and bad decisions on the setup. Is dockering, keeping the data handling itself outside of it and a few other easy (or not so much) precautions enough to have a somewhat smoother sailing?
Also, how much time do you need to keep things from failing apart?
In our system, the university libraries filled 90% of our textbook needs. Some books were highly sought after, especially physics and calculus, which were common for all STEM majors.
In those cases, we would run to the library first thing to get the books. If you missed out, someone would give you the PDF.
Professors would email the reading list before the first class with their recommendations, and even tell the students which libraries had each book. Other professors would have their notes and handbooks available on the website, and have some of the copy shops sell them for the cost of printing.
Just one note about the moon orbit around the Earth, it is far more subtle; almost just orbiting the Sun alongside Earth. I can't explain better than minute physics, highly recommend: youtube.com/watch?v=KBcxuM-qXec
There is no supply chain of baseballs and baseball bats in Brazil. That would be considered a "exotic" choice of sport, with those supplies only available at expensive stores with imported goods
I really like the technology and almost jumped in, but quickly realized the lack of the _network_ effect in a community.
I live in a dense urban area, and yet I see very few nodes nearby, where I image I will be left in starvation. Without a proper mesh, I couldn't send a message to a fried 2 miles away with no los.
I had a college at post grad that did something similar for music. Trained with some genre-specific songs in my country and had it generated a few thousand songs on that genre.
Then they had a LLM do a hierarchical monte carlo voting on the best songs, and made a album with AI generated melody and voice based on the generated lyrics.