A Wolfram Summer School 2024 Project
For this edition of the ‘AI for Science’ Newsletter, I share with you my work done under the guidance of the Wolfram Summer School, exploring how the wolfram language can be used to build semi autonomous LLM agents for assisting Scientists in preparing their data for AI and ML applications.
In the previous edition – “Where is the Data?”, I talked about the various sources and modalities of data that will be needed to build sophisticated AI model in Science, but one common challenge that prevails across the research data ecosystem is the preparation of this data to make it readily computable for machine learning. In my project, I explored how the combination of LLMs with Wolfram Language, can be a powerful tool in getting across this data hurdle. Here is the abstract for my project :
This project explores the integration of Large Language Models (LLMs) with the Wolfram Language stack to build semi-autonomous agents capable of performing diverse data science tasks. Leveraging Wolfram Language’s well-curated and computable knowledge base and functional development approach, this project demonstrates the use of LLMFunction and LLMTool as potential building blocks of agent like architectures. This work presents practical examples and future explorations towards developing autonomous workflow agents that generate Wolfram Language code for AI and data science applications. This could be a great tool to increase access to AI in the scientific community by simplifying data preparation and computability.
You can find the full project note-book (in a ‘computational essay’ form) over here : https://community.wolfram.com/groups/-/m/t/3209668
Future Explorations
Wolfram language provides a computational approach to build complex agent-like tools that could lift the data science burden off the scientists and allow them to easily prepare their data for downstream applications of AI and ML. I intend to explore the application of these and other tools to a broad range of scientific data and application areas and as always will share my findings and learnings in the coming editions.
About the Wolfram summer school
Attending the Wolfram Summer School was an exceptional experience, filled with invaluable learning and mentorship. The school’s rigorous and innovative environment pushed me to expand my understanding of AI and data science, particularly through the practical application of Wolfram Language. The guidance from the expert instructors and collaboration with fellow researchers were instrumental in advancing my project. I highly recommend this program to anyone eager to delve into the forefront of computational research. Moreover, I encourage everyone to explore the diverse and groundbreaking projects completed by my peers, which showcase the incredible potential of Wolfram technologies in various scientific domains.