Artificial intelligence (AI) as a novel tool to advance the field of medicine has rapidly become a subject of great interest and intense debate, leading many stakeholders to prospect on the future role of these technologies and relative responsibilities in decision-making [1–5]. Fur-thermore, the complexity of some AI algorithms, their lack of transparency and a widespread lack of prospec-tive validation may serve to dishearten physicians. In this short piece, I will attempt to define what are some of the most common AI techniques with applications to inten-sive care, then I’ll discuss what barriers to implementa-tion exist and why us, doctors, have not been superseded by robot-physicians.
The computer scientist Marvin Minsky, one of the fathers of AI, defined the field as “the science of mak-ing machines do things that would require intelligence if done by men”. AI has many potential applications in medicine, including automated radiology reporting, predictive analytics, knowledge representation, surgi-cal robotics, medical education, drug discovery, among others [1–5]. These technologies hold the promises of helping clinicians make better and more personalised decisions, boosting workflow and reducing healthcare expenditures [1, 3–5]. In the ICU, applications of AI are mainly limited to machine learning—a collection of data analysis and modelling techniques aiming at generating knowledge from data.
