Clinical AI at a crossroads: skill decay, robotic surgery, and the wearable data frontier

Surgeons remotely piloting humanoid robots completed laparoscopic cholecystectomies on a live pig, the first time such procedures have been performed by teleoperated humanoid systems, according to Forbes contributor John Koetsier. The milestone arrived the same week Google disclosed SensorFM, a foundation model purpose-built to interpret continuous wearable sensor streams, and a separate Forbes analysis by contributor Demetri Giannikopoulos raised pointed questions about whether heavy AI adoption is quietly degrading clinician skills. Taken together, the three developments frame a set of governance and procurement decisions that health system technology leaders can no longer defer.

Robotic surgery moves from the lab to the operating field

The humanoid robot surgery demonstration is significant not because unsupervised robotic procedures are imminent, but because it validates the underlying dexterity architecture. Koetsier's reporting in Forbes describes surgeons maintaining teleoperated control throughout a procedure that requires fine motor precision and real-time spatial judgment. That is a different class of capability than the fixed-arm surgical robots health systems have deployed over the past decade.

For surgical services directors and supply chain teams evaluating the next generation of robotic platforms, the question is less whether humanoid systems will enter procurement cycles and more when. The demonstration creates a new reference point for vendor conversations about dexterity, haptic feedback, and remote operation latency. Health systems that have not yet built an internal framework for evaluating emerging robotic surgery vendors will find themselves reactive when commercial options mature.

Google's SensorFM targets the wearable data gap

Google's SensorFM is a foundation model designed to process raw sensor output from wearables and surface clinically relevant signals, according to Forbes contributor Josipa Majic Predin. The distinction matters operationally: most wearable integrations today push device readings into an EHR or population health platform, leaving interpretation to care teams. A foundation model positioned at that translation layer could shift where the analytical work happens and, by extension, which vendors sit at the center of a health system's wearable strategy.

CIOs evaluating wearable programs should note that SensorFM is positioned as a platform-layer model, not a consumer product. That means its eventual integration points are likely to run through existing enterprise health IT infrastructure rather than around it. How Google distributes access, through APIs, cloud services, or partnerships with EHR vendors, will determine the procurement path.

Skill decay: the governance risk hiding in AI adoption

The concern Giannikopoulos raises in Forbes is precise: as AI systems handle more routine clinical reasoning, the frequency with which clinicians exercise independent judgment decreases, and with it their proficiency. The argument is not that AI produces wrong answers but that automation reduces the deliberate practice required to maintain diagnostic and procedural competency. This is a workforce and credentialing problem as much as a technology one.

Health system CNOs and CMOs who have deployed AI-assisted diagnostics, documentation tools, or clinical decision support without parallel competency monitoring programs now have a defined risk to address. That could mean structured case reviews that require unassisted reasoning, simulation-based credentialing, or audit mechanisms that track how often AI recommendations are accepted without clinical interrogation.

CIOs consolidating tech stacks as cyber risk climbs

The technology governance conversation is not limited to new AI models. Becker's Hospital Review reported this week that health system CIOs are actively working to rationalize tech stacks built up through years of layered point-solution acquisitions. The driver is partly operational complexity and partly security exposure: Becker's also reported that healthcare ransomware attacks rose 14%, a figure that makes every redundant or poorly integrated system a potential liability.

The timing is consequential. Health systems weighing new AI tools, whether for wearable data, surgical robotics, or clinical decision support, are doing so in an environment where IT leadership is simultaneously trying to reduce the number of active integrations, not expand them. Vendors that can demonstrate clean API compatibility with dominant EHR platforms and a credible security posture will have an advantage in that environment.

What this means for your team

• Audit current AI-assisted clinical tools for competency safeguards: determine whether your credentialing and simulation programs account for tasks now routinely offloaded to AI systems.
• Build a wearable data strategy before SensorFM-class models reach enterprise distribution: define which sensor streams feed clinical workflows and who owns the interpretation layer.
• Use the robotic surgery milestone as a vendor evaluation trigger: request roadmaps from current and prospective surgical robotics partners that address humanoid dexterity and remote operation capabilities.
• Tie any new AI or integration investment to tech stack rationalization goals: quantify the security and operational overhead of each addition before contracting.