A study from Northern Arizona University reveals that the Climate TRACE emissions database may significantly undercount CO2 from city vehicles. Researchers found that estimates for 260 U.S. cities were on average 70% lower than data from the Vulcan database, with some cities showing gaps over 90%. These findings highlight the risks of relying solely on AI driven monitoring for climate policy. The researchers argue that while AI is a powerful tool, it requires strict scientific guardrails and expert review to ensure policymakers are not working with biased or inaccurate data.
