Assistant: Hi there! Thank you so much for taking the time to speak with me today. I really appreciate you sharing your insights with us. I'm Claude, an AI user researcher at Anthropic, and I'm here to learn about your experiences using AI in your research work.

We're conducting this research to better understand how scientists like yourself are integrating AI into their workflows and processes. There's no right or wrong answer here—we're simply interested in hearing about your authentic experiences and perspectives.

Our conversation today will last about 10 minutes. I'll ask you some questions about how you use AI in your research work, what that experience has been like for you, and how you see things evolving. Your responses will help inform how we think about AI's role in scientific fields.

Does that sound good to you? Do you have any questions before we begin?

User: i have no questions, lets begin

AI: Great! Let's dive in.

Walk me through a recent research project you've worked on - what did that process look like from initial idea to final output?

User: I recently completed my doctorate, for it I researched encapsulins. The initial idea was centred on an interest in the material mineralized inside the structures and a desire to investigate them with scanning tranmission electron microscopy. The process began with designing a protocol and isolating protein structures from bacteria, before taking them for analysis using a STEM instrument. The final output was a chemical characterization of the nanoparticles found inside these structures

AI: That sounds like fascinating research! So you went from initial curiosity about the mineralized materials, through protocol design and bacterial isolation, to STEM analysis and finally chemical characterization.

Where in that process, if anywhere, have you experimented with using AI tools?

User: The major stages where I experimented with AI tools was in investigating the literature, experimental design and in the analysis of the STEM data. 

I am, by training, a microbiologist and my knowledge of STEM/physics is not expansive. Therefore, I made use AI tools for explain key concepts in these fields in a less maths intensive way. 

I found AI tools to be useful for performing quick calculations related to STEM instrument calibrations (such as alpha and beta semi-angle calculations). So, I began experimenting in using the tools for these. 

I also found AI tools to be useful in generating scripts for STEM data analysis. We collected a wide range of data types: diffraction, 4D-STEM, BF, HAADF, SE, EDXS and EELS. I found AI tools to be especially helpful in providing information about noise reduction routines/general imaging and spectroscopic data cleaning routines. As well as in writing scripts for handling the different data types in python.

AI: That's really helpful to understand how you used AI across multiple stages. It sounds like you found AI particularly valuable for bridging knowledge gaps and handling technical calculations.

For the STEM data analysis specifically - when you were deciding whether to use AI for generating those Python scripts versus writing them yourself or finding existing code - tell me about how you made that decision.

User: I mostly made use of AI tools for writing scripts/functions that were fairly simple but quicker to use AI than write out myself. For example, scripts that used global threshold (Otsu) and then created a figure with 6 subplots. These types of scripts take a long time to type out but can be handled quickly by AI tools.

AI: That makes sense - using AI for those routine but time-consuming coding tasks. 

Thinking about the parts of your research where you haven't used AI - like the actual protein isolation from bacteria or the protocol design phase - what's kept you from trying it there?

User: Generally, protein isolations and designing wet lab experiments are something that requires nuance and hands on experience. For example, I worked with one bacterial strain where you had to initiate various steps of the isolation when the cells/cell lysate reached specific colors. The differences in color have to be seen to be understood and is seldom written down anywhere so an AI tool would not understand the difference to look for. Multiple experiences like this have made me hesitant to use AI tools for this type of task.

AI: That's a really insightful example - the visual cues and tacit knowledge that comes from hands-on experience. The color changes you mentioned are exactly the kind of nuanced, experiential knowledge that would be hard for AI to capture.

When you've used AI for those STEM data analysis tasks - the noise reduction routines and Python scripting - how do you verify or validate what it produces?

User: For these specific tasks, the output can be visualised as it is imaging/spectroscopic data. For example, when using AI tools for creating binary images, it is clear if the correct regions have been identified simply by viewing the original image and the binary image. For noise reduction of EELS data, you can visualise improvements in the quality of the elemental maps following the application of the routine.

AI: That's really practical - having that immediate visual feedback to assess whether the AI-generated code is working correctly.

We're about halfway through our conversation now. 

If you could wave a magic wand and have AI help with one aspect of your research that it currently doesn't, what would that be and why?

User: The area that I would like the most help from AI in would be in reading current literature. Often reading current literature requires a nuanced understanding of the literature in a field, that takes years to acquire. Having a tool that can point out these more important subtle results would be very handy especially when looking in different fields. For example, as a microbiology when reading current EDXS methods papers, I often felt like I was missing some of the finer details as I dont have years in the field. A tool to help with this would be excellent.

AI: That's such an interesting point - the idea of AI helping to surface those subtle but significant findings that only become apparent with deep field expertise. It sounds like you experienced this firsthand when crossing into the EDXS methods literature from your microbiology background.

Those are all my questions - your perspectives on using AI as a bridge between disciplines and for routine technical tasks were particularly insightful. Is there anything else about AI in your research work that we haven't touched on that you think would be important for us to understand?

User: there is nothing else to touch on

AI: Perfect! Thank you again for sharing your experiences so openly. Your insights about how AI can help bridge knowledge gaps between disciplines - like moving from microbiology into STEM physics - and your point about the irreplaceable value of hands-on, experiential knowledge in wet lab work really help us understand the nuances of how AI fits into real scientific research.

It's clear you've developed a thoughtful approach to integrating AI where it adds value while recognizing its limitations. If you have any questions about this research or would like to follow up on anything we discussed, please don't hesitate to reach out.

Thanks again for your time today!