While I agree with the thrust of this answer, I don't think its is quite non-optional yet. No one in my group uses AI for, well, anything. I'd say only about 50% of the students in my bioinforamtics classes do. I use it occassionaly for explaining things to me, and have been thinking hooking up a code assistant to VS-code, but thus far everything i've done is hand coded.
I'm going back to do a master's in bioinformatics after four years away from academia, and i've been thinking about whether I should be suing AI to speed up.
I studied business analytics and then worked for 2 years and always use them for data, summarizing. Now that I‘m moving into bioinformatics, it feels like a completely different ecosystem. The AI toll and workflows seem much more specialized than what I used before.
So I'm trying to figure out what makes sense at this stage.
Should I use AI for my research or say what I should use?
2 answers
Use of AI is pretty much non-optional now (sadly).
Use LLMs to get you up to speed on new areas/techniques, generate code and generally improve your productivity.
Don't use it as a crutch to try and get by without really understanding what you are doing.
Vibe-coding might be a legitimate approach these days, but vibe-science or vibe-data analysis isn't.
Nobody will force you to use it, but I think if you aren't already using it, you're already falling behind (and may not realise it).
In particular, my answer is about using LLMs to do things like literature reviews, which I do on a regular basis. I don't actually use AI for any kind of data analysis or actual informatics at the moment (short of the occaisional AlphaFold run etc).
Yes, AI tools are very helpful for research, programming - especially in a language you don't know well, data analysis, plotting, building GUIs and more.
AI can help you bugfix weird errors with packages when building docker containers for example which can otherwise take a lot of time.
Log in to answer this question.
By the way, this blog post is really interesting. https://combine-lab.github.io/blog/2026/02/15/a-skeptics-guide-to-generative-ai-coding.html
I've seen top developers in bioinfo and more unknown devs creating really nicely documented, fully featured tools using AI coding agents. I've converted my own nextflow pipelines into bash with a single question (which actually worked, apart from some submission dependency modifications).
The productivity gains are real - especially for rapid prototyping.