My uni is using AI for assessment feedback
duti.dev
duti.dev
This doesn't really sound like AI to me, maybe machine translation or something like that but I never see a model making nonsense grammar mistakes and messing up tenses. Even a really small 7B model mostly outputs correct grammar until it goes off the rails. I'd expect bad AI feedback to be vague and verbose, in this instance perhaps whoever is writing the feedback doesn't know much English and is translating from their native language, or maybe they're outsourcing or something.
> "Will the application has a local repository of songs, will you use database to store songs?"
That question really doesn't make sense since it's specifically an Android development module and the MediaStore API dictates how files such as music are stored. It's definitely not the professor giving the feedback, that much is clear.
Of course there's an issue with how hard it would be to find a topic that AIs trained on the internet aren't already familiar with. So we'd need AIs to do research and create course materials for newly discovered science.
To get this working well, we'd need significant scale and would need to implement AIs acting as the university admins, AIs making decisions on grant allocation and AIs running peer-reviewed journals.
Eventually, applying similar approaches to other fields, we might entirely automate away humanity.