Mycin went the same way 30 years ago. It was more narrow, just recommending therapies for bacterial infections. It had three problems: 1) it was inconvenient to use, doctors had to examine the patient, consult with the computer, do the additional exam requested by the computer, consult with the computer again, then decide whether to take its recommendation 2) it was very brittle, if it got something within its knowledge base it easily surpassed the physician, if there was even something slightly outside it came to completely erroneous conclusions and seemed unable to tell when it didn't know the answer, so it gave, sometimes really bad advice 3) it threatened the physicians, who felt their role was being reduced to the eyes and ears of the machine.
Watson could have a vastly greater scope, but I have a hard time imagining that it wouldn't have the same three problems. Problem 2 would just occur less often. The biggest issue will be solving problem 1. It will take an enormous leap forward in user interface to make this usable. If the machine could act as a real time collaborator, that might work. It would require that the doctor and the machine be able to converse extremely reliably. It would also require that the machine be sensitive to the patient's needs. It would have to alert the doctor to certain possibilies without alarming the patient. It would also have to be sensitive to the needs of the doctor. It would need to keep track of what the doctor is doing reliably on his own and not remind him of things that he never misses. Put together this is a huge task in social engineering as well as computer science.
Problem 3, I think will solve itself. Students introduced to it in the classroom will be adopters, if the other two problems are solved. They will view the machine as "having their back". The machine will also empower them to deal competently with a wider range of diseases, because they will be able to broaden their procedural competence while the machine covers the diagnostic scope. Their successes will draw the older ones in. There will be die-hards, of course, but they will either be highly competent people that don't need the machine, or they will be edged out by their own flagging performance.
When they make a computer that turns on instantly, never forgets my email or printer settings and is immune to hacking and viruses I will be impressed. Watson is just an extension of current computing practice with some fancy programming.
As we design more and more complex machines, we become more and more confident in our ability to repair these machines. Ultimately we all become humbled when trying to either explain or listen to a set of symptoms regarding an illness of one of the most comlex creations in the universe. Good diagnosticians spend la ifetime studying every aspect of physiology and following research in the medical field and yet frequently mis-diagnose the simplest of maladies. Properly applied, Watson will be a tool to assist the physician in diagnosis and treatment. But ultimately, the decision will be in the hands of the Doctor. We should exercise due caution when empowering our health-care payers (not providers!) on how to determine what treatment is most appropriate. Given the current condition of our health care system and the organizations trying to run it, the concept of a "Watson" scares the heck out of me. We need a Watson, but we need to be very careful how we use it.
Medical applications do seem to be an ideal match for this type of advanced software technology. Lots of facts and data that can be analyzed and the need for advanced algorithms to quickly comb through large amounts of data. Will be interesting to see the "practical approaches" to using this kind of technology, especially given liability concerns and the need the absolute need for a personalized approach to medical diagnosis.
Medical is a great application for Watson. It will be interesting to see whether doctors will be willing to utilize Watson. It will also be interesting to see whether the insurance industry requires its use in order to reduce risk.
I agree that Watson is great for the initial research but I don't see computers replacing human intuition and experience in the trail stages.
This article brings to mind a lecture I heard that explored how humanity can't keep up with its own progress. We've found so many answers in the late 20th and early 21st century but we're not sure what the questions are. Using computers like Watson may get the right questions out there to lead to better advancement.
Geof, Watson is a great candidate for medical applications. For years, expert systems, of which Watson is an advanced example of, have been touted for medical applications. They have in fact proven themselves. The issue is liability. From my own experience, and that of others, we would all be better off if computers used more often in medical diagnosis.
New versions of BASF's Ecovio line are both compostable and designed for either injection molding or thermoforming. These combinations are becoming more common for the single-use bioplastics used in food service and food packaging applications, but are still not widely available.
For industrial control applications, or even a simple assembly line, that machine can go almost 24/7 without a break. But what happens when the task is a little more complex? That’s where the “smart” machine would come in. The smart machine is one that has some simple (or complex in some cases) processing capability to be able to adapt to changing conditions. Such machines are suited for a host of applications, including automotive, aerospace, defense, medical, computers and electronics, telecommunications, consumer goods, and so on. This radio show will show what’s possible with smart machines, and what tradeoffs need to be made to implement such a solution.