Narkosekrankenschwestern

I see CRNA as a likely candidate not to be replaced completely in the near term, but for AI to dramatically reduce the expertise needed. The complexity of administering anesthesia and the ease of making a mistake is exactly why it will be ripe for AI. AI already excels at making sense of complex data and reducing human error. "Anesthesia is an art form" Okay, well look at the art AI is already producing. It has already shown a propensity for creative writing, music, and images, among others. Oftentimes it's indistinguishable from human creativity. What I doubt AI can do the most is actually the dexterity portion of the job.

Step 1: Develop the AI. The data inputs are already there - AI can already read a patient's charts, monitor their vitals, and transcribe feedback/instruction from the surgeon. If an AI can be developed that takes that information and selects which drugs to administer, that removes nearly all the brainpower and expertise needed from a CRNA.
Step 2: Train someone to intubate and the other physical labor parts of the job. Pay them 1/3 of what you pay a CRNA.
Step 3: An Anesthesiologist monitors multiple rooms simultaneously (like they already do with CRNAs) as a failsafe against errors by the AI and is present during critical periods like induction and emergence, intervening if there's an emergency.

With the right machine they could administer anesthesia and monitor a patient simultaneously. Especially with the growth of engineering jobs in designing new technology for the medical field.

Anesthesia is one of the most complicated areas in medicine and is incredibly dangerous when even the slightest mishap occurs. It requires creativity as not every scenario is cut and dry. It is well beyond just "putting someone to sleep". Anesthesiology involves critical communication, ethical decision making, risk management, manual labor, emergency management, situational awareness, and knowing what to do when inevitably one of the many many medical devices we use fails. I believe one day AI will be able to accomplish any job but the sheer complexity of anesthesiology is far beyond the capabilities that the average person may think a robot will be able to achieve in 20 years.

we alredy use robot for micro stiches so we most likely will do that soon.

Every patient, and every case is different. No A.I. machine could handle all the nuances and multiple causative factors that determine the outcome of an anesthetic. Also, good anesthesia is part art and part experience and knowledge. A A.I. robot could never master this.

Manual dexterity, complex procedural skills, interpersonal skills, ability to comfort and reduce fear on part of the patient, rapid decision making based on knowledge, combined with intuition.