A forest notebook and a data centre may seem worlds apart. Conservation researchers are bringing them together, using artificial intelligence to make old observations easier to search. The harder question is how to judge the environmental value of that work alongside the resources needed to run it.
Reuters reporting published on November 7, 2026 highlights environmental organisations adopting AI despite concerns about data centres. One example is the Jane Goodall Institute's work with Amazon Web Services on records from Tanzania's Gombe National Park. The project is not a newly announced discovery today; its earlier documentation shows how the research has developed.
What AI does with a field notebook
Arizona State University described its contribution on April 8, 2026. Its researchers are digitising handwritten chimpanzee observations held in the Gombe archive. Computer vision straightens scans and extracts entries from structured field sheets, while language models help interpret handwriting and notes.
The university says students check the digitised results against the original sheets. That detail matters: converting a page into searchable data is not the same as proving that every interpretation is correct. A difficult symbol, a missing observation or an ambiguous time can still require a person who understands the record.
The Jane Goodall Institute's December 2, 2025 announcement described an AWS-supported research platform drawing on more than 65 years of primate behaviour records. That announcement establishes the project's history and intended scope, rather than confirming that every planned feature is operational now.
Faster research is useful, but it is not a climate balance sheet
Making records searchable can help scientists compare observations across years without repeatedly handling the paper originals. The practical test is whether the resulting dataset remains accurate, traceable and useful for a real conservation question.
A responsible assessment should ask what work was improved, how mistakes were measured and who can inspect the source record. It should also separate the project's computing footprint from the much larger footprint of the industry supplying it. Neither a promising demonstration nor a global electricity total establishes the net environmental effect of one research tool.
That distinction complements the habitat questions explored in our report on monarch migration and conservation: technology is a means of understanding living systems, not a substitute for protecting them.
Why the electricity question remains
In its 2026 report Key Questions on Energy and AI, the International Energy Agency says global data centre electricity demand grew 17% in 2025. Its central projection rises from 485 terawatt-hours in 2025 to about 950 in 2030. The latter is a projection, not a measured future outcome, and the totals cover data centres rather than conservation AI alone.
The IEA also distinguishes improving efficiency per task from rising overall demand. Wider use and more energy-intensive applications can increase consumption even when individual operations become cheaper to run.
What readers should look for next
For AsiaNews Iran, the useful follow-up is evidence: independent accuracy checks, documented research benefits and a transparent account of computing resources. Claims that AI is automatically green, or automatically useless to conservation, skip that work.
Which wildlife or environmental records would you most like to see made easier to study, and what safeguards would you expect?