By 2034, autonomous research fleets reconstruct changing deep-sea ecosystems from sparse sensor readings, enabling nations to study waters they can no longer afford to survey directly.
Research agents choose sampling routes, operate remote instruments, and infer ecological conditions across gaps between scattered observations. Human scientists increasingly set thresholds for disconfirmation, procedures for resolving conflicting results, and ecological red lines. Smaller nations gain access to deep-sea knowledge, but a persuasive reconstruction can still render an unobserved species an invisible casualty.
At 3:10 a.m. in a windowless laboratory in Cebu, marine ecologist Mara Santos compares two agent reconstructions of the same cold seep. One predicts a nursery habitat; the other, lifeless sediment. She spends the hours before dawn authorizing a single sampling dive whose result will determine whether twelve fishing boats remain in port.
A model that fills gaps in the evidence can also obscure how little was actually observed. Wealthy states may be able to fund more verification dives, while poorer states face pressure to accept model-generated certainty about their own waters.