A new artificial intelligence screening tool has demonstrated the potential to cut unnecessary glaucoma referrals by half while maintaining diagnostic accuracy comparable to human specialists. Published in The Lancet Primary Care, the study addresses the global challenge of glaucoma, a leading cause of irreversible blindness that often remains undetected until significant vision loss occurs. While population-wide screening has historically been deemed impractical, this research suggests AI could offer a viable, cost-effective solution for early detection within routine primary care. The investigation took place in 2023 at a single screening center in Lisbon, Portugal. Researchers screened 671 adults aged 55 to 65 using an AI system designed to analyze retinal images. For comparison, the same images were independently evaluated by six glaucoma experts. The results showed a significant divergence in referral rates: the AI system referred only 66 participants, representing 9.8% of the group, whereas the human experts referred 118 participants, or 18.0%. Despite the lower referral volume, the AI tool matched human performance in detecting the disease. It correctly identified 78% of individuals who actually had glaucoma, compared to 75% for the eye doctors. More notably, the AI demonstrated superior precision in ruling out the disease. It correctly excluded glaucoma in 95% of healthy participants, whereas human experts achieved a 91% exclusion rate. The study authors emphasize that this high accuracy in identifying those without the condition is critical. False alarms frequently lead to unnecessary hospital visits, increased patient anxiety, and added strain on healthcare systems. By minimizing these false positives, the AI tool could significantly reduce the burden on specialist services while ensuring that genuine cases are not missed. The authors propose that integrating AI-based screening into primary care could facilitate earlier detection and prevent avoidable vision loss. However, they also note specific limitations regarding the study's generalizability. The screening was conducted through an existing program for diabetic eye disease, meaning a large proportion of the participants had diabetes. Consequently, the observed glaucoma rates may not accurately reflect the prevalence in the general population without comorbidities. Despite this constraint, the findings offer promising evidence that AI can enhance the efficiency and effectiveness of glaucoma screening, potentially transforming how this sight-threatening condition is managed worldwide.