In just 48 hours, a Google AI has discovered what researchers spent ten years formulating. This could revolutionize scientific methods, redefine the role of researchers, and raise questions about our place in a future where artificial intelligence becomes a partner in discovery.
A decade of human research summarized by AI in one weekend
This scientific upheaval took place at Imperial College London. For ten years, a team has been trying to understand how certain bacteria manage to resist antibiotics. It was a monumental task, involving genetic analyses, clinical trials, and many sleepless nights filled with doubt.
Then came Co-Scientist, Google’s AI assistant based on Gemini 2.0, presented as a collaborative tool for researchers. Out of curiosity, the team submitted their topic to the AI. The result? It formulated the same hypothesis as the researchers… in just two days. Shockwaves rippled through the lab. In fact, one of the authors, José R. Penadés, confessed to having suspected Google of hacking, given how coincidental it was.
A tool capable of formulating hypotheses but not replacing experimentation
Of course, this achievement doesn’t mean that the AI could do everything in their place. As the team notes, proving a hypothesis scientifically takes time, rigor, testing, and failures. And that’s something Co-Scientist cannot fast-track.
However, what’s impressive is that if this hypothesis had been generated from the start, several years of trial and error could have been avoided. The AI also proposed four other plausible avenues, including one completely novel suggestion. As a result, this provides a boost to research… while giving the human ego a reality check.
An unprecedented analytical capability disrupting traditional methods
This isn’t the first time AI has been discussed as a catalyst for science, but this example marks a turning point. With their ability to assimilate and cross-reference millions of data points, these tools can formulate hypotheses or identify patterns that human brains would take months to detect.
Yes, they do hallucinate at times and make errors, just like us. Nevertheless, they free up time and cognitive space for researchers to focus on what truly matters: testing, understanding, and creating. José R. Penadés himself stated: “I believe this will change science, without a doubt.”
These AIs do not replace human thought, but they can detect weak signals within the sea of scientific publications, accelerate literature reviews, or even propose oversights. Thus, it represents a shift in pace, almost a paradigm change, transforming research into a more strategic endeavor than ever before.
Towards a new era of research: the complementarity between human and artificial intelligence
What if we stopped opposing researchers and AI? In this specific case, Google’s tool does not replace anyone. Instead, it acts as a highly efficient super-colleague that suggests, accelerates, and stimulates. The genuine brilliance still resides in the human hand, capable of validating, interpreting, and contextualizing these results.
In summary, the future of science may well look like a laboratory table where researchers and AI engage in dialogue, each contributing their strengths and weaknesses. And frankly, if that can prevent a decade of aimless wandering, why not take advantage of it?



