AI-Assisted Scientific Writing: From Literature Search to Publication
Dates
8-10 July 2025
To foster international participation, this course will be held online
Course overview
The rapid evolution of large language models (LLMs) and autonomous agents is transforming scientific writing — from literature search and synthesis to drafting, revision, and even peer review. This practical course provides researchers with hands-on experience in applying state-of-the-art AI tools to accelerate and enhance every stage of scientific writing.
Participants will explore the AI landscape, learn effective prompting strategies, and experiment with advanced multi-agent review workflows. Through guided exercises, they will build reproducible and auditable AI workflows, critically evaluate AI-generated content, and draft sections of their own papers using cutting-edge tools.
Target Audience
1. Researchers familiar with basic scientific writing in English
-
2. No coding background required
- 3. Free, browser-based tools (ChatGPT or Claude account recommended)
Learning outcomes
- Accelerate literature review and drafting using state-of-the-art LLMs and autonomous agents
- Design reproducible, auditable AI workflows for scientific writing and peer review
- Critically evaluate AI-generated content for accuracy, ethics, and transparency
Instructor resources provided to participants
- Slide deck summarising AI taxonomy, latest model capabilities and tool-selection heuristics.
- Colab notebooks for the Deep Research and Agentic Paper exercises.
- Pre-course survey template and post-course reference pack.
Session content
Day1 - 1-4 PM Berlin time
Focus: Evolution to transformer era, taxonomy of symbolic AI → generative AI, responsible use, data-privacy checkpoints
Hands-on components & key take-aways: Prompt lab: simple vs. JSON-structured prompts to build a manuscript outline
Day2 1-4 PM Berlin time
Focus: Prompt patterns for coherent sections, integrating domain data & statistics, collaborative workflows; overview of 2025 LLM advances (o3, Claude 4, Gemini 2.5, DeepSeek R1)
Hands-on components & key take-aways: Participants draft a section of their own paper; Deep Research demo – automatic search → reading → synthesis of hundreds of sources in minutes
Day3 1-4 PM Berlin time
Focus: Fact-checking, citation control, hallucination mitigation; multi-agent review concepts (MARG architecture)
Hands-on components & key take-aways: Guided run of Agentic Paper (https://github.com/albertogerli/Agentic_Paper): each attendee launches the pipeline on a short manuscript and analyses the specialised reviewers’ reports
COst overview
Cancellation Policy:
> 30 days before the start date = 30% cancellation fee
< 30 days before the start date= No Refund.
Physalia-courses cannot be held responsible for any travel fees, accommodation or other expenses incurred to you as a result of the cancellation.