In our EMBL Skills & Careers Webinar in October 2025, Jacobo Miranda, Bioinformatics Training Officer at the European Molecular Biology Laboratory (EMBL), explored how early career researchers can integrate artificial intelligence (AI) into practical aspects of their daily work, including idea development, project documentation, data security and choosing the right tools for writing and visualization. Jacobo shared hands-on examples and practical advice that speak to the common challenges researchers face when managing complex academic projects.

Below, you will find a short career profile of Jacobo and the full Q&A from the live webinar. Jacobo’s presentation slides and a summary of the key takeaways are also publicly available for your reference.

Career profile

Dr.-Ing. Jacobo Miranda is a Bioinformatics Training Officer in the Data Science Internal Support team at EMBL Heidelberg and a member of de.KCD, the German Competence Center for Cloud Technologies in Data Management and Processing. He completed his postdoctoral work as a research software engineer at the Max Planck Institute of Molecular Cell Biology and Genetics (MPI-CBG) and earned his PhD in Natural Language Processing, with a focus on machine learning and AI. His work centers on empowering researchers through training, support, and accessible online resources.

Webinar Q&A

You mentioned that you use AI regularly in your work. Are there still things that you prefer to do manually or in traditional project management software?

Oh yes, definitely. I develop a lot of training material, and I have a strategy for using AI to assist with that. However, I noticed that I lost track of what I truly knew when I let AI do too much of the work. For me, learning happens in the process of creating the material myself. So I went back to doing the core work manually and only used AI for parts I am less interested in, such as metadata or adapting quizzes to the audience. For my own learning, I need to do it myself.

Many of our attendees work with confidential project data. What are your main tips on what you can and cannot share with AI tools?

When in doubt, don’t share sensitive data. You can also run large language models (LLMs) on your own machine rather than sending data to the cloud. For example, LM Studio lets you download and run LLMs locally, keeping everything on your own device. If you need to handle larger batches securely, you can use a service like Chat AI on Academic Cloud, which offers a privacy‑focused environment for research-related workloads. These setups support researchers who want control over their data, rather than having it used to train external products.

Related to this, Claude AI, one of the examples you mentioned for privacy‑focused or data‑safe AI use, can users control which files it can access on their laptops?

Yes. when you enable Claude AI to access a file, it will always ask for your permission first. You can choose to allow or deny each request, so you remain in full control of what the AI can access.

How could AI help identify recurring issues or human errors in projects?

This is a very valuable question. When you are deeply involved in a project and focused on each next step, it can be difficult to step back and see broader patterns. By documenting all your steps and the issues you encounter, even by using voice-to-text or dictation, you can then ask AI to analyze the information and identify recurring patterns. It provides a different perspective that is hard to see when you are immersed in the work. So yes, make sure to document everything.

Do you have recommendations for AI tools that can take notes and turn unstructured discussions into structured documents, not necessarily on Zoom?

Yes. You can install software on your own computer that listens to meetings and transcribes them automatically. For generating summaries of text, I prefer ChatGPT for writing-focused tasks, and Perplexity for content that requires references. It’s worth trying different tools, as it only takes a few minutes to see what works best.

If you want to organize your notes and create a clear structure, I like NotebookLM. It lets you combine your own notes with material from other sources, and it includes a mind map feature that makes it easy to navigate key topics.

If someone shares an idea with ChatGPT to help expand and develop it, almost like brainstorming, is that acceptable?

Absolutely. AI has access to a lot of contextual information that you might not have or remember, and that is exactly its role. If it can help you develop your ideas, feel free to use it. Just remember that you remain responsible for the results, and it’s important to verify the suggestions against the actual literature.

What AI would you recommend for scientific visualizations?

I would not recommend asking AI to generate the visualization entirely on its own. Start by creating a rough sketch yourself, even a simple drawing, and then ask the AI to turn it into a polished SVG or image. I personally prefer Midjourney. Google also offers powerful image-processing tools, including one called Gemini 2.5 Flash Image.

Can you explain how using hashtags (#) helps generate metadata and organize documents?

I use a tool called Obsidian, which works with Markdown files. Instead of placing files strictly inside folders, I add hashtags within the files, for example, #ProjectA and #ProjectB. This allows a file to belong to multiple categories at once. Tools like Obsidian can then sort and display the files as if they were in those folders.

You can also add metadata, such as when the file was created, who created it, and what it relates to. This additional context helps AI understand and select the relevant information when responding. Using hashtags and metadata together makes documents easier to organize and more accessible for AI.

Finally, what AI tools would you recommend for academics, especially for writing papers and literature reviews?

There are language models designed specifically for writing academic papers, but I do not recommend using them to generate entire papers. AI should be used to support and enhance your own work, assisting you piece by piece rather than replacing your writing. I don’t have any specific tool recommendations for this.

Additional resources from the EMBL Fellows’ Complementary Skills Programme are available on our website.