“By double-checking, by validation, and referring to the original references, an IP can configure that and know it,” Molaeb said.
Kamangu argued that this makes foundational IPC knowledge more, not less, important.
“You cannot use an AI tool and be 100% proficient with it if you don't have the foundation,” she said. Without that expertise, an IP may accept an incorrect AI-generated answer as fact, potentially creating patient safety and cost consequences.
“We need to keep using our creativity, our critical thinking as humans, and make sure we know the basics and the foundation of the work,” Kamangu added.
From Surveillance to PPE Training
Molaeb described several areas where AI could complement infection prevention work. Surveillance systems can analyze laboratory results, microbiology data, device utilization, antimicrobial use, and clinical documentation to flag unusual patterns for review.
AI can also support education through simulations and computer vision. Molaeb described tools used to train health care personnel in hand hygiene and PPE donning and doffing.
If a staff member misses a step, “AI can indicate that,” he explained. Traditional one-to-one observation may not consistently capture every movement, whereas AI can make errors immediately visible, allowing correction “before the practice reaches the patient care and could do a patient harm.”
Environmental services represents another opportunity. AI-enabled monitoring may identify frequently missed high-touch surfaces or variations in cleaning practices. But workload, room design, patient acuity, and workflow can influence those patterns, meaning human interpretation remains necessary.
“The IP's role will give the context, will do the validation, and enhance the final decision-making,” Molaeb said.
AI Literacy Is Becoming an IPC Competency
Avoiding AI altogether may create its own risks. Molaeb warned that organizations restricting sanctioned AI tools could unintentionally encourage “shadow AI,” in which employees independently use unapproved applications.
Without basic AI literacy, staff could share sensitive information, construct ineffective prompts, or accept inaccurate results.
“IPs, they do not need to become data scientists and experts in AI,” Molaeb said. “But they need to understand how to question, how to build a prompt, how to validate, and how to use AI tools safely.”
Successful implementation also cannot belong solely to IT. Molaeb called for collaboration involving IP leaders, frontline users, IT, clinical informatics, quality and patient safety, privacy, cybersecurity, and executive leadership.
“If the end user is not involved in the process, it's really hard to implement and scale,” Kamangu added, particularly when health care organizations are handling sensitive patient and enterprise information.
AI Could Strengthen the IP's Seat at the Table
Rather than diminishing the IP's role, Molaeb believes AI proficiency could increase infection prevention's influence within health care organizations.
Competencies in data interpretation, validation, governance, and prompt engineering can help IPs generate faster analyses and stronger evidence for decisions.
“AI may automate parts of the surveillance,” he said, “but it cannot replace the IP's clinical context, their leadership, or accountability for prevention.”
His final message returned to the reassuring premise of his APIC presentation.
“The goal is not simply to adopt AI,” Molaeb said, “but to use it safely, responsibly, and meaningfully to improve patient and health care worker safety.”
And, he added, perhaps IPs really can “take it easy, because artificial intelligence is here, and it will be our new best friend in our practice.”
