credit: Andrew D. Czaja An AI system identified previously undetectable patterns in 3.3-billion-year-old rocks, revealing the strongest early evidence for life on Earth, where Carnegie Science researchers’ use of AI in life sciences discovered a shift in inquiry tools from human-led hypothesis to AI-enabled revelations.
In a study of South African sediments, the finding used AI to analyze volumes of datasets and discern complex, non-linear patterns invisible to the human eye. AI, acting as a leading research partner, is now deployed to accelerate discoveries across different fields, from genomics to material science to cosmology.
The study’s finding surface as time when some of the biggest AI developers, including OpenAI, publicly predict AI’s impact will migrate from consumer chatbots to drivers of constructive scientific discoveries within the next three to five years.
AI and Biological Discovery
Earth formed approximately 4.5 billion years ago, and scientists believe life began about 3.7 billion years ago. Until now, direct biochemical signs only went back to 1.6 billion years. The researchers said the new method was a step forward for AI and R&D, changing the way scientists handled extremely large datasets.
In the research, published in Proceedings of the National Academy of Sciences, scientists found microbial “fingerprints” in 3.3-billion-year-old rocks and signs of photosynthesis in rocks 2.5 billion years old. It identifies minimal chemical features that even trained experts cannot see, helped partly by advances in machine learning for science.
“Scientists have developed many different ways to infer life in ancient samples… but using complex molecules to come up with an unambiguous record of life only extended previously to about 1.6 billion years ago,” co-lead author Michael L. Wong said.
AI in life sciences relies on 3D spectral readings, an approach reinforced in developments in AI physics research wherein similar tools spot patterns in massive datasets.
Researcher Prabhu compared the system to facial recognition and said its accuracy is increasing sharply, benefiting from techniques seen in deep learning in molecular biology. This technology could help recover biomarkers even from damaged rocks, an effort supported by the growing field of biological and AI discovery.
New Forms of Life Detection
AI in scientific research now regularly uncovers patterns difficult for humans to detect complex chemical data.
The researchers highlighted the power of AI science discovery that can expose deep geological samples’ evidence. It is the beginning of new early-life analysis possibilities, empowered by advances in that use AI for science in endless disciplines.
Better AI scientific data visualization is improving the ability to scan ancient samples with high precision, making it easier to interpret hidden features during analysis. Wong believes the success of the system on Earth encourages the use of AI in new science missions to Mars or icy moons.
He added that the model was lightweight and fast, which supports real-time field decisions and falls in line with new methods for AI hypothesis testing used in modern labs. Given the training on GC-MS data, this model could also support work in AI for materials science, which depends on similar chemical readouts.
The team using AI for science expects future versions to be even more capable with more available data and to further strengthen the role of AI in life sciences in research.
Researchers now hope to team up with NASA in developing space-ready versions, which would extend AI in science beyond Earth. For now, this fast-developing industry is rewriting the story of how life originated and may soon help reveal whether it exists elsewhere.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Intelligent Tech sections to stay informed and up-to-date with our daily articles.