The healthcare industry treats artificial intelligence like just another software rollout. We write policies, form committees, and assume standard guardrails will keep operations safe. But agentic AI behaves entirely differently than previous digital health tools.<\/p>\n
At the recent eHealth26<\/a> conference in Halifax, Nova Scotia, I sat down with Julia Zarb. She is the force behind Blue x Blue<\/a>, a company building AI infrastructure with governance incorporated. She warned that applying the old IT playbook to this new medium creates massive unseen risks.<\/p>\n Core Insight: Human Oversight is a Liability Trap Slapping a \u201chuman-in-the-loop\u201d label onto an AI tool does not guarantee patient safety, it simply shifts the legal and operational liability onto overwhelmed clinicians.<\/p>\n The Human-in-the-Loop Fallacy<\/p>\n Executives love hearing that a human will review AI outputs before finalizing a clinical decision. In reality, these reviewers lack the time and system explainability to catch microscopic errors under high stress.<\/p>\n \u201cWhat we\u2019re doing is transferring the liability that comes with that review,\u201d Zarb noted. \u201cWe don\u2019t have what\u2019s called explainability to go back in and see how that got to it, who knew what, when, and where, and why?\u201d<\/p>\n The Threat of Invisible Errors<\/p>\n AI models fail in tiny, logical steps that easily bypass human reviewers.<\/p>\n \u201cWhat makes agentic AI good is also its internal flaw,\u201d Zarb explained. \u201cThe way that it processes, it carries forward little micro fissures and flaws.\u201d<\/p>\n Standard IT guardrails fail because AI naturally jumps boundaries. When multiple AI agents interact, a single logical error propagates until it results in a much larger failure.<\/p>\n \u201cIf 100 agents speak to 100 other agents and those mistakes are brought forward, these aren\u2019t mistakes that are just errors, these are mistakes that make sense [to Ai at the time] because of the way the AI is built.\u201d<\/p>\n Questions Healthcare IT Leaders are Asking<\/p>\n Why is the \u201chuman-in-the-loop\u201d approach risky for healthcare AI? Relying solely on human oversight transfers immense liability to clinical staff. Fast-paced environments and high stress prevent humans from properly auditing AI outputs. Furthermore, clinicians lack the technical explainability tools required to understand exactly how the AI arrived at its conclusions.<\/p>\n How do AI errors differ from standard digital health software bugs? Traditional software bugs are often clear and trackable. AI generates \u201cmicro fissures\u201d or small gap-fillers that look perfectly logical to the system. These tiny errors bypass standard guardrails and multiply rapidly when different AI agents share data across an organization.<\/p>\n Learn more about Blue x Blue at https:\/\/www.bluexblue.com\/<\/a><\/p>\n Listen and subscribe to the\u00a0Healthcare IT Today Interviews Podcast<\/a>\u00a0to hear all the latest insights from experts in healthcare IT.<\/p>\n And for an exclusive look at our top stories,\u00a0subscribe to our newsletter<\/a>\u00a0and\u00a0YouTube<\/a>.<\/p>\n Tell us what you think. Contact us\u00a0here<\/a>\u00a0or on Twitter at\u00a0@hcitoday<\/a>. And if you\u2019re interested in advertising with us,\u00a0check out our various\u00a0advertising packages<\/a>\u00a0and request our\u00a0Media Kit<\/a>.<\/p>\n Get Fresh Healthcare & IT Stories Delivered Daily<\/p>\n Join thousands of your healthcare & HealthIT peers who subscribe to our daily newsletter.<\/p>\n We respect your privacy and will never sell or give out your contact information<\/p>\n","protected":false},"excerpt":{"rendered":"The healthcare industry treats artificial intelligence like just another software rollout. We write policies, form committees, and assume…\n","protected":false},"author":2,"featured_media":122975,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[179,7493,1798,32086,4378,17366,62058,4989,62059,34607,62060,2609,62061,6975,62062,4855],"class_list":["post-122974","post","type-post","status-publish","format-standard","has-post-thumbnail","category-agentic-ai","tag-agentic-ai","tag-agentic-artificial-intelligence","tag-ai-governance","tag-ai-guardrails","tag-ai-in-healthcare","tag-ai-liability","tag-ai-risk-in-healthcare","tag-ai-safety","tag-blue-x-blue","tag-clinical-ai","tag-clinical-workflow","tag-digital-health","tag-ehealth26","tag-human-in-the-loop","tag-julia-zarb","tag-patient-safety"],"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/122974","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=122974"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/122974\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/122975"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=122974"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=122974"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=122974"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}