Massachusetts Institute of Technology (MIT) researchers recently announced that artificial intelligence (AI) has successfully designed two novel potential antibiotics capable of killing drug-resistant gonorrhea bacteria and methicillin-resistant Staphylococcus aureus (MRSA). The research team stated that these two drugs were designed by AI at the atomic level and successfully eliminated superbugs in laboratory and animal experiments. However, they still require years of refinement and clinical trials before being ready for clinical use.

Currently, antibiotic resistance poses an increasingly serious problem, causing over one million deaths annually. Overuse of antibiotics has prompted bacteria to evolve mechanisms to evade drug effects, while development of new antibiotics has remained stagnant for extended periods.

Previously, researchers used AI to screen known chemical substances to identify potential new antibiotics. The MIT team has taken this approach further by directly utilizing generative AI to design antibiotics targeting gonorrhea and the potentially lethal MRSA.

The team's research findings have been published in Cell journal. They trained AI to analyze 36 million compounds, including those not yet discovered. Researchers provided the AI with chemical structures of known compounds and data on their effects on different bacterial growth. The AI learned how different molecular structures affect bacteria to design new antibiotics.

During the design process, the AI also excluded structures too similar to existing antibiotics and ensured the designs were drugs rather than soap-like substances, while eliminating compounds toxic to humans.

Researchers tested the AI-designed antibiotics against gonorrhea and MRSA through laboratory bacterial testing and infected mouse experiments, ultimately obtaining two novel potential drugs.

MIT Professor James Collins stated: "We are excited because this demonstrates that generative AI can be used to design entirely new antibiotics. AI enables us to create new molecules in a low-cost, rapid manner, thereby expanding our arsenal and giving us an advantage in the battle of wits against superbugs."

However, these drugs are not yet ready for clinical trials and are expected to require one to two years of improvement before beginning the lengthy human testing process.

Dr. Andrew Edwards from the Fleming Initiative and Imperial College London considers this work "very important" with "enormous potential" because it "demonstrates new methods for identifying novel antibiotics." However, he also noted: "While AI promises to significantly improve drug discovery and development, we still face arduous efforts in testing safety and efficacy."

This process is lengthy and costly, with no guarantee that experimental drugs will ultimately reach clinical use. Some experts call for further improvements in AI drug discovery technology.

Professor Collins stated: "We need better models that not only consider how drugs perform in the laboratory but also better predict their effectiveness in the human body."

Additionally, the manufacturing difficulty of AI-designed drugs presents another challenge. Although AI theoretically designed 80 top treatment options (compounds) for gonorrhea, only 2 could be successfully synthesized and manufactured into drugs in practice.

Professor Chris Dowson from the University of Warwick considers this research "cool," indicating that AI has made "significant progress" in antibiotic discovery and helps address resistance issues. However, he also pointed out that drug-resistant infections present an economic problem - "How do you manufacture drugs with no commercial value?"

If a new antibiotic is invented, ideally it should be used sparingly to maintain its effectiveness, but this makes it difficult for pharmaceutical companies to profit from it.