The problem Learn from AIchemists solves

Our project “Learn from AIchemists” attempts to create interactive educational course, where an enthusiast can paste his desired lectures link (YouTube Playlist) or videos. The proposed system, will achieve the following

Parse through all the videos in the course playlist.

Summarizes the content by exploiting SOTA Speech to text algorithms. The summary is generated in such a way that prior expertise is not required to understand the text.

AI Teacher Assistant, is available which can answer your queries from the course

AI Quizmaster is embedded into the system which can generate questions from the specific lecture based on the desired level of difficulty. Additionally, this assesses the student as well.

There are immense practical implications in real life, primarily in Educational Sector. Many students use lectures hosted by NPTEL , MIT-OCW, Standford Online, and other MOOCs. This platform aids the students by answering their queries automatically and testing their learnings from the course.

Education is the most important and significant expenditure in an individual’s life in this era. Though this facility is free in few countries, but huge number of human resources are needed. By exploiting the use of existing open-sourced SOTA AI systems, we have built this prototype. Although our system has the capability to query ChatGPT [5] we have also incorporated its alternative for each task.

Video lectures date back to more than 30 years, there exist huge amount video archives which easily covers up-to Master level courses. We believe with necessary architecture our system can be scaled up to a stage where one completely designs AI-certified degree program, keeping aside government approvals which is a challenge out of the scope of this event.

Challenges we ran into

Handling long videos and large texts

Computational limitation

Limited access to existing solution API keys, such as ChatGPT