AI is Rewriting Evolution – Signs of Life Found in 3 Billion Old Rocks

Scientists may have just uncovered evidence that life began on Earth nearly one billion years earlier than we thought — and artificial intelligence is the reason we’re finally seeing it.

The finding challenges long-standing timelines of when biology first took hold on our planet and suggests early photosynthetic life may have appeared far sooner than textbooks assume.

And if these results hold up, the implications ripple far beyond Earth.

AI Detects Biological “Echoes” Hidden for Billions of Years

Specifically, the research in question combines modern geochemistry with machine learning to reveal ‘chemical echoes’ of early organisms. These echoes survive even when heat and pressure have long erased traditional fossils, according to the study published in PNAS on November 17.

As it happens, researchers trained AI models on more than 400 samples, including modern plants, animals, fossil microbes, meteorites, and ancient sedimentary rocks. By learning how biological molecules fragment differently from non-biological ones, the model could distinguish the two with high accuracy, even when the original biomolecules were destroyed.

When applied to rocks older than three billion years, the system detected strong biosignatures, including evidence of photosynthetic organisms. This pushes back the timeline for oxygen-producing metabolism by nearly a billion years.

According to Michigan State University scientist Katie Maloney, an assistant professor in the Department of Earth and Environmental Sciences, who contributed rare billion-year-old seaweed fossils, the method reveals biological clues that were previously ‘invisible,’ allowing researchers to “read the deep time fossil record in a new way.”

The study also showed that microbial-like patterns appear consistently in Paleoarchean samples, strengthening earlier findings based on morphology and isotopes.

Why This Discovery Could Redefine the Search for Alien Life

If life appeared this early on Earth, it suggests two major possibilities: life emerges quickly when conditions allow, or Earth was unusually fertile during its earliest era. Either scenario has major implications for planetary science and astrobiology.

The method can also separate biological organic matter from meteorite-derived carbon, a critical distinction for studying rocks from Mars, Europa, Enceladus, or comet samples. Future space missions often hinge on identifying whether complex carbon signatures came from living organisms or from abiotic chemistry.

Researchers plan to expand the technique by integrating isotope data, Raman spectra, and infrared signatures, potentially unlocking even deeper insights into early life and ancient metabolisms. According to Dr. Robert Hazen, senior staff scientist at Carnegie Institution for Science and a co-lead author:

“Ancient life leaves more than fossils; it leaves chemical echoes. (…) Using machine learning, we can now reliably interpret these echoes for the first time.”

If true, the findings paint a picture of a planet teeming with microbial activity long before complex organisms and far earlier than standard models predicted.

More Must-Reads:

• 3I/ATLAS Leaks Nickel Gas Far From the Sun – NASA Can’t Explain It
• Kissing Didn’t Start With Humans – Research Says It Began 21 Million Years Earlier
• Groundbreaking Discovery Unearths 39,000-Year RNA of Woolly Mammoth