Genoa, December 07, 2024 – The first artificial intelligence specialized in Earth Sciences has been developed: it can date rocks, analyze paleoclimate, and interpret landscapes. This significant milestone is described in the study titled “Can AI Get a Degree in Geoscience?”, recently published in the prestigious scientific journal Geoheritage (https://doi.org/10.1007/s12371-024-01011-2). The research results from the work of geologists Andrea Baucon (University of Genoa) and Carlos Neto de Carvalho (Naturtejo UNESCO Global Geopark), who developed the AI to tackle complex geological problems.
The new artificial intelligence, GeologyOracle, has been developed to fill a gap. “Generative AI systems have shown extraordinary capabilities,” states Andrea Baucon, “but they lacked solutions specifically tailored to Earth Sciences.” Thanks to an innovative training approach, GeologyOracle offers users the opportunity to interact as if they were communicating with a human geologist. For example, a user can ask, “Why did the dinosaurs go extinct?” and receive a detailed response in natural language.
Can GeologyOracle earn a degree in geology? Baucon states, “We are almost there.” An international committee of university professors has evaluated the performance of this new artificial intelligence by subjecting it to 152 Earth Science questions, mimicking a very rigorous university examination. GeologyOracle demonstrated notable competence, correctly answering 79.6% of the questions and achieving numerous ’10/10′ scores (the highest marks). “Despite the excellent results, I recommend a critical approach: artificial intelligence can make mistakes,” explains Neto de Carvalho.
GeologyOracle can assist scientists in understanding the dynamics of our planet, acting as a scientific assistant capable of suggesting new hypotheses and interpretations. For instance, this innovative system can analyze temperatures over the past 100 million years, identifying patterns and causes of climate fluctuations. Thanks to its multimodal nature, GeologyOracle accepts input in the form of text, audio, and images. For example, to date a rock, one simply needs to take a photo of its fossil content, and GeologyOracle will provide a scientific interpretation, estimating the sample age in millions of years.
The launch of GeologyOracle is scheduled for November 14 on https://geologyoracle.com/, in honor of Charles Lyell’s birthday. Andrea Baucon states, “It is a tribute to one of the fathers of modern geology,” emphasizing how Lyell has been an inspiration for the development of this tool. The new artificial intelligence stands out as a versatile resource, designed not only for researchers in the field of geosciences but also for students, educators, and geotourists. These users will have free access to GeologyOracle, which aims to contribute to the dissemination and accessibility of geological knowledge.
Can AI Get a Degree in Geoscience? Performance Analysis of a GPT-Based Artificial Intelligence System Trained for Earth Science (GeologyOracle).
Baucon, A., de Carvalho, C.N. Geoheritage 16, 121 (2024). https://doi.org/10.1007/s12371-024-01011-2
| TITOLO | DOI | RIVISTA | ARGOMENTI DI RICERCA |
| Can AI Get a Degree in Geoscience? Performance Analysis of a GPT-Based Artificial Intelligence System Trained for Earth Science (GeologyOracle) | https://doi.org/10.1007/s12371-024-01011-2 | Geoheritage | Geologia, informatica |
Lo studio ha beneficiato della collaborazione di istituzioni scientifiche italiane (Università di Genova) e portoghesi (Naturtejo UNESCO Global Geopark).
GeologyOracle will be released on November 14, 2024, at the following address:

Andrea Baucon (left) and Carlos Neto de Carvalho (right) during fieldwork at Naturtejo UNESCO Global Geopark (Portogallo).

Andrea Baucon engaged in field data collection at the Penha Garcia palaeontological site within the Naturtejo UNESCO Global Geopark.

Andrea Baucon engaged in field data collection at the Penha Garcia palaeontological site within the Naturtejo UNESCO Global Geopark.

Ichnofossil photographed during field data collection at the Penha Garcia palaeontological site within the Naturtejo UNESCO Global Geopark.

The new artificial intelligence can correctly identify fossils like this trilobite (collections of DISTAV – University of Genoa).

The new artificial intelligence can correctly identify fossils like these ammonites (collections of DISTAV – University of Genoa).

GeologyOracle banner.

GeologyOracle logo.