AI WORLD HQ
If you've ever taken a course of antibiotics for a sinus infection or a scraped knee, you've used one of the miracles of modern medicine. Most of the antibiotics your doctor prescribes were discovered more than 60 years ago. Since then, bacteria have gotten smarter — and many infections are now harder to kill. Finding new antibiotics has been painfully slow.
That's starting to change, and AI is one reason why.
Why new antibiotics are so hard to find
Antibiotics are substances that kill bacteria. Most of them originally came from nature — from soil microbes, fungi, and other living things that produce them as a defense. The traditional way to find a new one was to grow millions of microbes in petri dishes and test each one to see if it stopped bacteria from growing. Imagine tasting every spoonful from a giant pot of soup to find the one with the perfect seasoning. That's roughly how slow the process has been.
On top of that, many of the most promising "sources" — animals that went extinct thousands of years ago, or microbes that live in places we can't easily reach — were almost impossible to study. We knew the genetic material (the DNA instructions inside every living thing) was sitting in museums and databases, but nobody had time to read it carefully.
How AI changes the search
This is where tools like Codex and ChatGPT come in. Codex is an AI made by OpenAI that writes and understands computer code. Researchers use it to write scripts — small programs that can scan huge DNA databases in hours instead of years. ChatGPT, the chatbot you may have used to write an email or summarize a long article, helps scientists brainstorm, summarize research papers, and explain patterns they find.
Put simply: a researcher can now ask, "Which of these 10 million DNA sequences look like they might produce a substance that kills bacteria?" — and get a short list of likely candidates to test in the lab. The AI doesn't replace the lab work. It just makes sure the lab doesn't waste months on dead ends.
In one notable project, a research group led by scientist César de la Fuente has used exactly this approach — combining AI analysis with lab testing — to find new antibiotic candidates hidden in genomes from both living organisms and extinct ones (like woolly mammoths). Early results have turned up a few molecules that show real promise against drug-resistant bacteria.
What "drug-resistant" actually means
You may have heard the phrase superbug in the news. It refers to bacteria that no longer respond to common antibiotics. Things like MRSA (a stubborn staph infection) or certain tuberculosis strains fall into this category. When an antibiotic stops working, simple infections can become life-threatening again. Doctors call this antimicrobial resistance, or AMR.
The World Health Organization has called AMR one of the biggest threats to global health. New antibiotics are the main way we fight back, but the discovery pipeline has been mostly empty for decades. Anything that speeds it up — even a little — is a big deal.
Wrap-up
The search for new antibiotics used to be a slow, expensive grind. AI tools are now helping researchers scan millions of genetic sequences in days instead of years — and turning up real candidates worth testing in the lab. None of this changes your doctor's appointment next week, but it's quietly reshaping how we prepare for the next generation of infections. If you want to follow this kind of work, look up César de la Fuente's lab or search "AI drug discovery" on PubMed — you'll find new papers every month.