Latin inscriptions from ancient times provide insights into the proclamations of the Roman emperor the voices of the enslaved for those who understand them. Today, advancements in artificial intelligence assist historians in piecing together often incomplete texts, even allowing for predictions about the origins and contexts of specific inscriptions within the Roman Empire.
“Analyzing history through inscriptions resembles solving an enormous puzzle. There are tens of thousands of them,” stated Thea Sommerschield during a media briefing at the University of Nottingham, UK. “This is everything that has endured through the centuries.”
The AI tools developed by Sommerschield and her team can identify missing characters in Latin inscriptions and indicate that they relate to the same individual or location as similarly styled inscriptions. They named the tool Aeneas, after the legendary hero reputed to have fled the fall of Troy, eventually becoming an ancestor to the Romans.
“Aeneas can help restore unknown lengths of text that are missing,” remarked Giannis Assael, co-leader of Aeneas’s development at Google DeepMind. “This versatility is beneficial for historians, particularly with heavily damaged materials.”
The team trained Aeneas using the most extensive dataset of ancient Latin texts, comprising over 176,000 inscriptions and nearly 9,000 accompanying images. This training enables Aeneas to propose missing phrases. Furthermore, when tested against a set of known inscriptions, researchers found that Aeneas could estimate inscription dates within a 13-year range and achieved a 72% accuracy rate in discerning the origins of different regions in Rome.
“Inscriptions are crucial for understanding the lives and experiences of those in the Roman Empire,” noted Charlotte Tapman, a researcher at the University of Exeter, UK, who was not directly involved in this study. “They encompass a variety of topics including law, trade, military affairs, political life, religion, death, and personal matters.”
Such AI technologies “are poised to be implemented for studying inscriptions from other eras and adapted for various languages,” according to Tupman.
In tests involving intentionally damaged inscriptions, Aeneas achieved a 73% accuracy in filling gaps of up to 10 Latin characters. However, when the total length of missing text was unknown, accuracy dropped to 58%. Nonetheless, the AI’s logic for its suggestions enables researchers to evaluate the legitimacy of its findings.
When nearly 20 historians evaluated AI-assisted restorations on intentionally damaged inscriptions, those using AI outperformed both the individual historians and the AI itself. Furthermore, historians reported that the comparative inscriptions highlighted by Aeneas served as valuable starting points for 90% of their research.
“Utilizing this technology accelerates work related to inscriptions, especially when seeking to build broader analyses of local or empire-wide patterns and epigraphic traditions,” stated Elizabeth Meyer from the University of Virginia. “However, it remains essential for human scholars to examine the results and ensure their plausibility within the given historical context.”
“Employing a generic AI model to assist in ancient historical tasks often yields unsatisfactory results,” commented Chiara Senati from the University of Vienna, Austria. “Thus, the creation of tools specifically catered to Latin epigraphy research is greatly valued.”
The ultimate vision is for historians to be able to “redirect their energies towards museums and archaeological sites,” Sommerschield shared during a press event.
IT is a warehouse resembling the size of 12 football pitches, poised to provide essential employment and development opportunities in the city of Caucaia, northeastern Brazil. Yet, the shelves remain empty. This extensive facility is set to transform into a data center, as designated by TikTok, になったんです。 English: The first thing you can do is to find the best one to do. part of a 5.5 billion Reais (7.3 billion pounds) project aimed at expanding the global data center infrastructure.
With the increasing demand for supercomputer facilities, Brazil is attracting an array of high-tech companies, buoyed by the AI boom. The selection of Caucaia is strategic. Submarine cables carry data from Fortaleza, the nearby capital of Ceará, to various continents. Proximity to these cables enhances traffic capacity and reduces latency—the response time across the Internet network.
Additionally, Caucaia is home to the Pecém EPZ, where businesses can produce goods and services for export, benefiting from various tax incentives and streamlined bureaucratic processes.
However, data from Brazil’s disaster digital atlas and integrated disaster information system indicate that Caucaia is also prone to extreme weather events, including drought and heavy rainfall.
Between 2003 and 2024, the city experienced drought-related emergency conditions declared at least once. In 2019, around 10,000 residents were impacted by water shortages. The digital atlas of disasters shows that as reservoirs depleted, water quality diminished, leading to crop failures and challenges in access to basic food supplies.
Data centers consume vast amounts of energy and water to keep supercomputers cool. Nevertheless, public agencies are promoting green construction in drought-affected areas. Caucaia is part of a broader trend.
According to the Digital Disaster Atlas, five of the 22 planned data centers are situated in cities that have faced repeated drought and water scarcity since 2003.
So far, China’s social networks have not been mentioned in Caucaia’s permit application. However, in February, the chief of staff for the state government, Chagas Vieira, confirmed in an interview with local radio stations that discussions were ongoing with Chinese firms, and representatives from TikTok and its parent company ByteDance met with senior officials, including the Vice President and Minister of Development, Industry, Trade, and Services, Geraldo Alckmin.
ByteDance has been approached for comments.
The truck will deliver water to Caucaia, a city facing repeated problems with drinking water supply. Photo: Marília Camelo/The Guardian
The project is officially led by Casa dos Ventos, a Brazilian wind energy firm that has invested in the data center sector. Mario Araripe, the company’s founder and president, announced last year that he aims to attract major global technology companies like Apple, Amazon, Google, Meta, and Microsoft to fill the facility with computers.
Casa dos Ventos has already secured one of the three required licenses from the state of Ceará. According to the state’s Environmental Supervision (SEMACE), the project received a license for “30m³/day water consumption in closed circuits supplied by Artesian Well.” Specific details have been withheld for commercial confidentiality.
Casa dos Ventos claims it is “committed to transforming Porto do Pecém into a complex of technological innovation and energy transition.”
Projects requiring significant energy, such as data centers, are required to obtain special permission from the Brazilian government. As of 2024, at least seven of the 21 approvals granted by the Ministry of Mines and the Ministry of Energy were linked to data centers.
Casa dos Ventos is also responsible for another data center project currently under state review in Campo Redondo, Rio Grande do Norte, a region that has experienced drought for 14 out of the last 21 years. During the water crisis in 2022, local governments sought federal aid, and water trucks were dispatched to address the demand.
A similar situation is unfolding in Igaporanga, Bahia, where a Brazilian renewable energy company plans to establish two data centers. The city has been in a state of emergency due to drought conditions for 12 of the years between 2003 and 2022. In 2021, about 5,500 people faced rural water shortages.
Transparency regarding water usage by data centers under construction in these areas is lacking. Companies have not disclosed this information voluntarily, and the government has withheld technical documents for licensing, citing commercial confidentiality.
In early April, the National Electric Power System Operator (ONS) denied requests for access to the grid for Casa dos Ventos, citing concerns for grid stability. Consequently, the Ministry of Mines and Energy requested a recalculation to assess potential grid adjustments.
bIG tech firms acknowledge their water consumption in water-scarce areas heightened by AI requirements. The 2024 Sustainability Report details that Microsoft reported that 42% of its water usage originates in regions experiencing water stress. Similarly, Google stated that in the same year, 15% of its water consumption fell in areas marked by “high water scarcity.”
Data centers utilize a large volume of water to prevent overheating in computers and machines. However, some water may evaporate, potentially exacerbating the local climate crisis where they are located. As AI technologies evolve, the demand for processing power increases, leading to heightened energy and cooling requirements. Consequently, water and energy consumption are projected to rise.
Workers at a Data Center in Porto Alegre, Rio Grande do Sul, Brazil. Such facilities utilize considerable amounts of water for cooling machinery. Photo: Jeff Botega
The International Energy Agency projects a significant increase in data center energy consumption to double, reaching 945,000 GWh by 2030—equivalent to Japan’s annual energy consumption. Countries like Brazil will account for approximately 5% of this growth within that timeframe.
Water consumption is expected to surge. Researchers from the University of California, Riverside, and the University of Texas at Arlington estimate that global AI demand will require between 4.2 billion and 6.6 billion cubic meters of water by 2027, surpassing half of the UK’s annual water usage.
However, Shaolei Ren, a researcher from UC Riverside and co-author of the study, highlights a crucial distinction between consumption (water extracted from the system) and loss (water evaporated).
“Residential users generally do not withdraw significant amounts of water, but data centers often consume between 60% and 80%,” notes Ren, meaning that much water is lost.
Data centers can be cooled through two approaches: one is air conditioning, a widely adopted method for various facilities, while the second is utilizing water.
The outskirts of Caucaia, where inadequate water became unsuitable for urban consumption after reservoirs were depleted in 2019. Photo: Marília Camelo/The Guardian
One method involves recycling or reusing water but incorporates fans and radiators within closed systems, resembling car engine technologies. Alternatively, a cooling tower might use evaporation to expel heat from heated water, allowing the return of cold water to the system. The final method involves misting water into the air, increasing humidity and reducing temperature.
Nonetheless, these methods are not without inefficiencies. “Both evaporation and misting lead to water loss,” asserts Emilio Franceschini, an associate professor at ABC Federal University.
A small data center with a capacity of 1MW consumes around 25.5 million liters of water annually, with an estimated 1% (255,000 liters) lost to evaporation.
In Pecém, alternatives to extracting water include purchasing desalinated seawater or recycled water from Fortaleza.
It falls upon the state government to grant water concessions to data centers as part of the environmental licensing process.
rOnildo Mastroianni, technical director at Esplar, an NGO with a 50-year presence in Ceará, argues that projects demanding high water consumption in semi-arid areas are misguided. “It’s simply pushing for increased dryness,” he asserts.
Mastroianni cautions that such projects could alter the local hydrological basin, which may weaken fragile ecosystems, like the Caatinga, and heighten food insecurity due to rural water scarcity. He indicates that representatives from local NGOs and various Kilombola and Indigenous communities were not included in project discussions.
Due to water stress, many communities have constructed reservoirs to secure water supply during drought periods. Photo: Marília Camelo/The Guardian
Other Latin American nations are also witnessing a surge in the data center industry. Chile has launched 22 data centers in the Santiago region alone. In December, the government announced a National Plan to establish 30 additional projects, projected to place the country at medium to high levels of water stress by 2040, signifying decreased water availability.
In Chile, both governmental and corporate bodies are facing escalating opposition. In 2019, Google disclosed plans for its second data center in Santiago, which sparked estimates from the activist organization MOSACAT indicating the project would extract 700 million liters of water annually.
Following a wave of protests, a Santiago court reviewed the project. By early 2024, the court halted Google’s assessments concerning environmental impacts, pending further evaluation.
Among those advocating against the project was Tania Rodriguez of MOSACAT, who lamented, “That turned into extractivism,” she said in interviews with other outlets. “We will become everyone’s backyards.”
Leonardo Augustus Patterson was born on April 15, 1942 in Limon, the Caribbean coastal town of Costa Rica. Little is known about his family history. He said his father left home when he was very young and his farmer mother died when he was a teenager.
He said he found his first ancient, ancient pottery shard in the Yam field when he was seven years old.
He moved to San Jose, the capital of Costa Rica, when he was 15 years old. There he found a job with a jeweller.
After selling items locally for several years, he moved to Miami, where he imported them to a local gallery. A chance encounter with a wealthy collector led to an opportunity in New York, where he had moved by the late 1970s.
Patterson kept his family’s life a lot secret. He claimed that five women had at least 13 children. The list of survivors was not available immediately.
Few people suspected that Patterson had committed a long list of crimes, but even some of his opponents have admitted that he has a loving side. They cited his gentle personality and his dry wit – a quiet charisma that appears to take on his entire career.
“He was adorable guy,” Brand said.
When he told Patterson he was planning to write a book about him, Brand recalled, and Patterson replied, “Wait until you’re dead.”
Latin America has been a source of inspiration for various aspects, including a popular literary and musical genre and staple foods like potatoes. A famous Happy meal is now an indication of this inspiration. There is potential for Latin America to also become a cradle for AI.
A coalition of research institutes is collaborating on a project called latamgpt, which aims to create a tool that considers regional language differences, cultural experiences, and “specificity.” This tool is intended to provide more accurate representations for users in Latin America and the Caribbean compared to existing Large Language Models (LLM) primarily trained by US or Chinese companies in English.
The project lead, Rodrigo Duran Rojas, expressed the importance of developing local AI solutions to better serve Latin America. The goal is to offer a representative outlook tailored for the region, with initial tests showing promising results in areas like South American history.
Over 30 institutions are involved in the development of Latamgpt from countries across the hemisphere, including collaborations with Latinos in the US like Freddy Vilci Meneseth, an associate professor of Hispanic Studies at Lewis & Clark College, Oregon.
Latamgpt’s launch is planned for around June, following a significant commitment from various regions for improved AI governance. Projects like monitoring deforestation in the Amazon Rainforest and preserving historical documents from past dictatorships are contributing to the dataset used for training Latamgpt.
With a dataset of over 8 terabytes, Latamgpt aims to provide a nuanced and localized model for various applications. The project faces challenges in incorporating diverse dialects and complex grammatical structures, but emphasizes the importance of collaboration for continued development.
Diversified dialects and complex grammar challenges
Efforts like Latamgpt, CHATGPT, and Google’s Gemini are working towards incorporating a wider range of data and improving localization for non-English languages. Challenges in training models for languages with complex grammar and dialects persist.
Despite these challenges, Latamgpt aims to address these issues through collaboration with institutions, libraries, and archives across the region. The project continues to receive data and feedback to enhance its capabilities and explore applications in public policy and regulation.
The long-term goal of Latamgpt is to create an interconnected network for developing AI solutions with a Latinx touch, emphasizing the impact of collaboration in shaping the future of technology in Latin America and beyond.
This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.
Strictly Necessary Cookies
Strictly Necessary Cookie should be enabled at all times so that we can save your preferences for cookie settings.
