

Papers GPT
Research

Jesse Zhang is a software engineer and entrepreneur who founded Lowkey, a gaming startup that was acquired by Niantic in 2021. He now works on the social team at Niantic and is also involved in some funds via Sequoia Scouts and Neo. He has a background in engineering (primarily web and ML) and math, and he studied Computer Science at Harvard. He also likes playing basketball, tennis, board games, and video games (he hit Grandmaster in Teamfight Tactics). He also wrote a problem on the 2021 AMC 10. He has a personal website where he showcases some of his cool projects and experiments with various technologies.
In this blog post, I will give you a brief overview of how PapersGPT works and what kinds of questions you can ask it. I will also show you some examples of the data sources and the answers that PapersGPT can generate. If you want to learn more about PapersGPT or Jesse Zhang, you can visit his website at https://jessezhang.org/llmdemo or follow him on Twitter at @thejessezhang.
Pros
- It is a demo of feeding various data sources into GPT to give it access to deep customized knowledge.
- It supports different types of data, such as scientific papers, Google Docs, GitHub repos, etc.
- It has a terminal-themed Q&A interface that allows users to ask questions about the data source and get answers from GPT.
- It is a creative and educational project that showcases the potential of GPT and natural language processing.
Cons
- It requires users to have their own OpenAI API key, which may not be easy or affordable for everyone.
- It may not be able to answer all questions accurately or comprehensively, depending on the quality and quantity of the data source and the limitations of GPT.
- It may pose some ethical or legal issues if users feed sensitive or copyrighted data into GPT and share the answers publicly.
- It may not work well on mobile devices or smaller screens due to the design and layout of the interface.