  The Stanford One Hundred Year Study on Artificial Intelligence (Stanford University, 2016) talks about various uses of AI in various fields as well as entry barriers or other obstacles that could hinder the uses of AI in those fields. For example, under the health care system, various broad claims were made with caveats, such as gAI-based applications could improve health outcomes and quality of life for millions of people in the coming years, but only if they gain the trust of doctors, nurses, and patientsh, and other problems such as gA small group of companies control the EHR market, and user interfaces are widely considered substandard, including annoying pop-ups that physicians routinely dismissh. The problems listed particularly under healthcare analytics only seem to list (1) gthe FDA has been slow to approve innovative diagnostic softwareh, gHIPAA requirements for protecting patient privacy create legal barriersh, gAI research and innovation in healthcare are hampered by the lack of widely accepted methods and standards for privacy protectionh, garchives have only gone digital over the past decadeh, and only lastly after all these points is listed almost as an afterthought, gvalue proposition is not yet compellingh. I argue that the current biggest barrier to entry of AI into the medical field is not any of the items listed previously, which do indeed pose larger downstream effects, but rather a fault on the shared on side of the computers themselves: the cost. Current medical facilities want to use this technology but the cost is currently too high.

  The other day, a family member of mine was talking to their doctor, and the doctor said something that stuck with me. If I remember correctly, he said that he is very interested and eager to use the new AI technology, but its currently too expensive to use practically. There are many software available from small startup companies or large companies like IBM, but the most famous one is probably IBM Watson. IBM Watson is a supercomputer. The supercomputer is not something that can be easily just installed to a laptop in the same way that one could install google chrome. Rather, what I infer to be the situation is that one must pay a lot of money to gain the licensing fee first, and then a lot more money to rent out enough space/time on a supercomputer to start making dents in patientsf cases. I think that currently, we are just simply too ahead of our time for this sort of technology to gain momentum as quickly as the authors seem to be hoping for. We need to wait a few more years until the quality of computers gets better and thus the price to use a super computer goes lower. That still wonft do anything to solve the licensing issue, but at the current rate, I daresay that getting a diagnosis from the computer may cost more than a diagnosis from the doctor.

  Thus, the AI industry gets too easily swept up in its own broad claims, blaming the lack of trust on the side of the doctors as well as strict regulations from the government for a slow start in acceptance and usage, where they themselves should share some of that burden.
