Seminar - Who does the science in the age of agents?

Seminar - Who does the science in the age of agents?

How agentic AI alters the academic landscape.

---

The fast development of large language models (LLMs) in recent years is starting to drive scientific discovery. Large-language-model agents are evolving from interactive assistants to full scientific operators: they write code, inspect data, launch analyses, design new methods, and draft reports. In this seminar we explore how agentic AI is altering the academic landscape with a focus on bioinformatics/biomedical research and have a closer look at the historic innovations that sparked this development. We divide the seminar into 5 broad categories:

1.  Basic concepts
2.  Multi-agent labs
3.  Lab in the loop systems
4.  Risks and Limitations

 

Organization:

• Wednesday 10:00-12:00, first presentation 04.11.2026 
• Kick-off: 21.10.2026, 10:00

Attendance is mandatory!

 

Formalities:

Presentations are 30 minutes, followed by a 30-minute discussion. Students will be evaluated on how well they present their paper(s), answer questions, and participate in the discussion rounds (including asking questions themselves).

This course is aimed at Master's students in Bioinformatics, Computer Science, and DSAI. Prior knowledge of Machine Learning or a related lecture is recommended.

Bachelor students may also register for the seminar (as a proseminar worth 5 CP), but only a limited number of slots are reserved for them.

 

Time Table:

After the kick-off meeting on the 21st October, the course will take place weekly from 4th of November onward. In each session we will listen and discuss the presentation of one student to guarantee enough time for Q&A sessions. 

 

Registration:

To register for the seminar, please sign up via the CMS by 19.10.2026, 23:59 CET. If your study program requires additional registration steps elsewhere, please make sure to complete those as well. We cannot be responsible for missed registrations outside the CMS.

During the registration, you will be asked to give your preferences for the 12 topics listed below (topic 1 = tutorial 1). Please indicate at least two topics with "preferred" and one with "okay". 
Topics can cover one or several papers, and you can find a description for every topic in the materials section.

After the registration deadline, we will review capacity and let you know whether you have been accepted or placed on the waiting list.
Attendance at the kick-off meeting is mandatory. Missing it without prior notice will result in exclusion from the course, and your spot will be given to a student on the waiting list.

 

Topic List:

1.  From Local to Global: A Graph RAG Approach to Query-Focused Summarization
2.  ReAct: Synergizing reasoning and acting in language models
3.  Co-Scientist
4.  Applications in Biotech: The Virtual Biotech Lab
5.  Spatial Agent
6.  Claude Science
7.  Physical AI
8.  Lab in the loop for perturbation design
9.  Can an agent replace a PhD? End-to-end automation
10.  Multi-agents failures
11.  Security risks of agentic system
12.  Dual use in biomedical agentic AI