Vibing with your research: The future of vibe research for biology

Science is in its most exciting era but with funding cuts and lack of resources, researchers are struggling to keep up. Every day brings a stream of new papers, data, and ideas, and in the middle of this pressure to do more with less, a new way of working is starting to emerge. It’s called vibe research, and it could reshape how biological research is carried out.

How can research be a vibe?

Born from the already popular “vibe coding” (a term coined by Canadian Computer Scientist Andrej Karpathy to describe using AI to write working code), “vibe research” is where AI agents will do the heavy lifting in the lab. A scientist sets a problem, then the AI reads papers, generates ideas, runs analyses, and writes up findings to help work through that problem. The idea is that you don’t have to be a coding genius to write code, and you don’t need to be an] expert in every facet of science to use AI for research.

With vibe research, the scientist still leads, but the AI handles the heavy lifting: scanning and summarising content quickly; highlighting insights, and making connections we might miss. It’s not about replacing scientists, it’s about freeing them to focus on the things that really matter, like understanding the overall picture, while AI handles the routine work.

Agentic AI is paving the way for vibe research 

Tech companies and academic researchers are already starting to develop agentic AI that supports this approach. Owkin’s K Navigator, Google’s AI co-scientist, and Stanford’s Biomni, are the most recent examples of agentic co-pilots that act as partners that help scientists talk to their data as easily as they talk to a colleague. 

Instead of simply assisting with individual tasks, AI agents already coordinate and run entire research workflows with a surprising degree of autonomy. Owkin’s K Navigator is already demonstrating its ability to multiply researcher productivity by as much as 20 times, giving scientists back time, energy, so they can focus their creativity entirely on their research. 

What makes vibe research different is how independently the agentic AI systems will operate. In the future, they won’t just respond to instructions, they’ll decide, for example, to run a follow-up experiment if a result is unclear, or to pull in additional sources when drafting reports. 

Traditional research is human-led, deep-research is AI-assisted, and vibe-research is AI-led with human oversight. It doesn’t cut people out, it shifts their role so that researchers can focus on the big questions that drive discoveries. Imagine every researcher across the world suddenly gifted a tireless lab assistant.

Human thinking and creativity is critical  

While vibe research opens up exciting possibilities, it also raises concerns that the AI agents may hallucinate facts, spread bias, and compromise data privacy. Getting it right will require a mix of technical safeguards, clear policies, and strong human oversight.

Technical fixes like retrieval-augmented generation, hallucination detection, and reproducibility logs can reduce errors, while policy measures could enforce transparency, authorship standards, and ethical safeguards (for example regarding AI model bias). And although this is a huge time save for researchers, they will still need to keep their analytical skills sharp by questioning results and checking for bias and areas the AI’s thinking could be improved. 

Agentic AI should help scientists, not replace them. With good design, smart rules, and people staying actively involved, we can tap into the power of vibe research while keeping the integrity and creativity that science depends on.