We Gave Robots The Steadiest Hands In Medicine – Now What?

Surgical robots have been in operating theatres for decades, but for most of that time they’ve been tools – highly capable tools, but ultimately extensions of a surgeon’s hands rather than autonomous actors. However, a new wave of systems is taking on procedures that require a level of steady-handedness most humans just can’t pull off – like retinal surgery, endovascular work or sub-millimetre suturing. In these areas, early results are showing something interesting: robots aren’t just keeping up with surgeons; in some cases, they’re doing better.

These questions go well beyond the technology. When a system makes a split-second, high-stakes choice during a procedure, where does the accountability land? How do we handle medical training when machines start outperforming human hands in those core manual skills? And there’s the big reality check: will patients actually buy into this, and will it ever reach beyond the best-funded hospitals?

What’s Ready Now

The best argument for automation is found in procedures where the extra precision and consistency make a clear, measurable difference for the patient.

Microsurgery and supermicrosurgery are the most obvious places to start. In these fields, robots can effortlessly smooth out the human tremor and fatigue that build up over a long operation, keeping the sub-millimetre precision required for nerves, vessels and lymphatic work. This isn’t just theory – the tech is already being used in operating rooms today.

Another natural fit for this is retinal surgery. The surgical margin in retinal work is extremely narrow and the consequences of a slip are severe. Recent clinical validation work has described soft robotic systems designed specifically for retinal procedures that operate close to or beyond what physiological hand limits allow. In standardised endovascular procedures, an early first-in-human study of automatic robotic-assisted aortic repair reported 100% technical and clinical success across four patients – a small data set, but a promising indicator.

What these cases have in common is structure: they involve tasks that can be broken into defined, repeatable steps, in anatomical environments that are reasonably predictable. That’s the condition under which automation performs best. The more a procedure relies on unpredictable anatomy, split-second judgment, or the kind of improvisation you only get from reading tissue and bleeding patterns in real time, the less ready it is for a robot to handle alone.

The Training Gap: Teaching Surgeons To Take The Wheel

The transition from surgeon-as-operator to supervisor brings significant challenges for medical training programmes. Robotic systems reduce tremor, filter fatigue and improve consistency – those are true benefits. But they also risk narrowing the hands-on exposure that trainees need to develop the manual competence to intervene when things go wrong.

If trainee surgeons spend their formative years supervising robotic execution rather than performing procedures manually, the surgical workforce that emerges will be skilled at managing automated systems but potentially under-prepared for the cases where those systems can’t complete the job. In practice, the medtech startups building these systems and the hospitals deploying them haven’t agreed on what that boundary looks like.

The emerging answer from surgical training researchers is that future surgeons will need a different skill profile: more competence in imaging interpretation, robotic interfaces and workflow management, alongside the escalation judgement to decide when to convert from automated to manual. Whether current training programmes are being redesigned to reflect that is unresolved in most health systems.

Who Is Responsible When It Goes Wrong?

Liability remains the most difficult unanswered question in surgical automation, and it’s a hurdle the legal system has yet to clear.

In conventional surgery, accountability for poor outcomes rests primarily with the surgeon, though hospitals and device manufacturers can also play a role. When a robotic system takes an active role in decisions, assigning clear responsibility becomes significantly more complicated.

Academic reviews of autonomous surgical systems explicitly frame this as a liability problem. The decision-making is distributed across the surgeon, the hospital, the software, the manufacturer and the procurement chain – and the contribution of each to any given outcome is difficult to isolate after the fact. This complexity is manageable while the robot is purely assistive, but it becomes a challenge the moment the system starts making autonomous choices without direct surgeon authorisation.

Patients are often told robotic surgery means more precision and faster recovery – which is true – but rarely told in any detail what degree of automation is involved in their specific procedure. The option to decline robotic assistance in favour of fully manual surgery exists in principle, but in practice it’s rarely presented as a genuine choice. As systems become more autonomous, informed consent needs to catch up with the engineering.

Does This Reach Everyone, Or Just The Best-Funded Hospitals?

The NHS has formal procurement routes for surgical robots through NHS Supply Chain and NHS Shared Business Services framework agreements. In the UK, surgical robotics is no longer just a trend; it’s now a recognised procurement category with set purchasing protocols. The challenge, however, is that access to this technology remains deeply unequal, dictated by hospital budgets, patient numbers, and available expertise.

The most likely trajectory is that automation will first enhance care in elite centres, then diffuse unevenly based on each system’s ability to absorb capital costs and maintain the necessary throughput. For public health systems facing budget constraints, this raises a familiar question: does the technology expand access or concentrate it? At current price points, surgical robotics risks the latter – at least until competition or greater scale eventually brings costs down.

The technology is advancing faster than the accountability structures designed to govern it, the training models built to prepare for it and the procurement processes needed to distribute it equitably. That isn’t an argument against the technology; it’s an argument for ensuring our policy and ethical discussions keep pace with the engineering.