AI가 약사를 대체할 것인가?

Pharmacy work falls into two broad kinds. The first is getting the right product into the right bottle: entering the prescription, counting, labeling, checking that the product matches, and managing inventory. The second is deciding whether the medicine is right for this patient. That means checking the dose against kidney function, spotting interactions, counseling, adjusting therapy and working with prescribers. Automation has been moving into the first kind for decades. Hospitals have used automated dispensing cabinets and carousel systems for a long time, and large-volume robots and central-fill sites now do much of the retail counting. Computer vision and barcode scanning have made product checks faster. The second kind is harder to automate. It needs context the software often lacks, such as what the patient actually takes, what they can afford and what they are willing to do. It also carries legal accountability. In the United States, state boards of pharmacy regulate who may verify and dispense, and a licensed pharmacist remains responsible for that work. Some states allow tech-check-tech, where trained technicians verify other technicians' fills in defined settings. That shifts checking from pharmacists to technicians, not to software. A common misconception is that AI explains retail pharmacy closures and staffing strain. Those pressures come mainly from reimbursement economics, including low dispensing margins and pharmacy benefit manager contracts, plus workload and burnout. Large language models can draft drug information, answer questions and summarize charts. However, they can produce confident errors, so their output needs a pharmacist's review. The profession is repositioning toward clinical work. Examples include residencies and board certification, ambulatory care clinics, collaborative practice agreements, test-and-treat services, pharmacogenomics, and advocacy for recognition as healthcare providers who can be paid for clinical services. Where those roles are funded, a pharmacist's value depends less on how many prescriptions they fill.