A new study from Blue Cross Blue Shield reports that hospital use of AI tools added $942 million to healthcare spending over two years. That number should stop us cold, not because AI is failing, but because we are deploying it in ways that add cost without demanding better outcomes. This is not a story about technology being bad. It is a story about how we measure value, and who benefits when we skip that step.
We have talked before about the gap between what AI promises and what it actually delivers. When our editor tested an interactive AI clone, the experience left her questioning the technology rather than trusting it. That is the same tension playing out in hospitals right now. AI is being used to document visits, predict readmissions, and support coding decisions. Those are useful tasks, but they do not automatically improve care. If a tool makes a billing process faster or a note more complete, the cost shows up on the books while the clinical benefit stays invisible. The Blue Cross data suggests we are paying for convenience, not health. And convenience, it turns out, has a steep price tag.
For our readers, the practical takeaway is blunt: do not confuse adoption with progress. Just because a hospital says it uses AI does not mean the technology is saving money or saving lives. We need to ask sharper questions about what those tools are actually doing. Are they reducing redundant tests? Are they catching errors that would have harmed someone? Or are they just adding a layer of administrative polish? The same logic applies to how individuals evaluate AI tools in their own work. We have argued that you need to verify an AI's understanding before trusting it, especially in high-stakes settings like tax preparation. Healthcare is no different. If a hospital cannot show that its AI investments lead to measurable improvements in patient outcomes, then the $942 million is not a sign of innovation. It is a sign of unchecked spending.
What we would tell a reader who asks about this study is simple: watch the incentives. Hospitals are not adopting AI because it is cheap. They are adopting it because vendors sell it as a way to streamline operations, and because the fear of falling behind is real. But the Blue Cross findings reveal a different reality. When the cost lands on insurers, and eventually on premiums, the burden shifts to patients. The question is not whether AI belongs in medicine. It does. The question is whether we are willing to demand evidence of value before we write the checks. That means asking for transparent metrics on how AI affects care quality, not just workflow speed. And it means holding vendors accountable for outcomes, not just features.
The concrete point to watch is this: if the $942 million figure grows, expect insurers to push back. They will start denying claims that rely on AI-documented services, or they will demand discounts from hospitals that cannot prove their tools work. That pressure will force a reckoning. Some hospitals will adapt and show real gains. Others will quietly retire their tools. The ones that thrive will be those that treat AI as a clinical tool, not a cost center. For the rest, the bill will keep rising, and we will all be paying it. The study is not a verdict on AI. It is a warning about what happens when we let the technology lead without asking who it serves.
