proper rock
**Project Name or Use-case: **Teacher Writers Assistant
**OpenAI Technology Utilized: **ChatGPT & GPT 3.5/4.0 API
**Project Details: **
Generating assignments for students with different levels of expertise en-masse with custom assignment configurations and instructions.
Background description
This project came to life with the OAI API, which ended up saving me tons of time: I had a problem with time management;
I have to guide and teach dozens of people every week and was manually writing out assignments and collecting information documents for them.
That took up around 10-12 hours a week for me, time I couldn't spend with my students or I'd have to overwork.
I also have a max writing speed so I could only write 4-5 assignments per week which comes down to 1 assignment per programming language. That was and is obviously insufficient.
That is where ChatGPT comes in to test out the capabilities and to prompt-engineer the perfect prompt for this usecase, so that we can then use it as a system instruction via the API.
Once the prompt was finalized, I wrote out a total of 92 assignment "configurations" in a spreadsheet : a title, small summary, requirements, language and what expertise level should be expected from the reader.
This allowed me to write a small app that iterates over all those assignment configs and query the API for them, after which i proofread and corrected where i saw fit.
Summarizing
By developing an app for this usecase I'm now able to save 10+ hours per week and spend it on something more important (like actually teaching).
Generating 92 assignments was under 17$ back when I developed this. Had i needed to write out all the assignments manually, it would've most likely taken me weeks.
When we calculate 10 hours per week's wage versus the costs of generating 92 assignments via the AI and proofreading them; then the later is the better option.
sadly i only have raw dutch input/output.
crimson wigeon