What is the problem?

LLM-based research assistants are designed to support scientists throughout the entire research cycle, from idea generation -> finding and understanding related work -> experiment design -> experiment execution -> data analysis -> publication.

At the Max Planck Institute for Polymer Physics, an experiment to calculate the frictional force in a liquid drop is being automated. Based on this experiment and field of research, this thesis aims to determine whether LLM-based research tools can already be of help in research in the natural sciences.

The project involves the following aspects:

• What LLM-based research assistants are available?
• What are their strengths and weaknesses?
• Test 2-3 assistants in a case study.
• Improve the assistants with respect to experiments in physics (Master thesis)

What do I need?

• Skills in Python
• Willingness to engage with LLM-based systems
• Knowledge of physics is not necessary but the willingness to learn the basics of the underlying physics

Literatur:

Blog post about an AI co-scientis

Nora from the Allen Institute for Artificial Intelligence (unpublished)

Here a quote from

“”“ The second is on an unreleased project called Nora, a research assistant agent for scientists. Not only can you interact with it (e.g., chat and ask questions), but it promises to execute code, understand literature, provide topic summarization, and more. Farhadi clarifies that it’s intentionally different from tools like GitHub Copilot or Augment Code, as their goal is to broaden its scientific scope. They plan to expand into new fields, including life sciences, starting next year. ”“”