Watson Who?

There is already a thread regarding Watson on Jeopardy. Interested indivduals can get to it from here.

I would rather comment here, because the wording of the blog is a little different.

What catches my attention here is the "spectacularly wrong" comment. This may go together with yet another blog about medical diagnosis by Watson-like computers.

If the offered medical diagnosis would follow this pattern displayed by Watson (either correct or spectacularly wrong), then it may be a viable idea. The quadratic equation offers two solutions, one correct and one spectacularly wrong. When the equation is solved, it is up to the solution seeker to discard the incorrect result. Expanding on this, a medical diagnosis that is spectacularly wrong should not be difficult to identify.

I follow your logic. However, it's only spectacularly wrong to someone who has some idea of what's going on. For instance, most people who travel via airlines in the US are aware of Chicago and would have known that Toronto was not the correct answer. However, if you are not all that familiar with the airline industry you may default to Watson. Some say that Toronto is an absurd answer because it's not in the USA....it's in Canada. However, there is a Toronto in Iowa and there is a Naples in Florida, and an Athens in Georgia.

And then there was this poor son-of-a-gun:

http://articles.latimes.com/1985-04-03/local/me-28522_1_europe-vacation

I would certainly hope that our microchip enhanced diagnostician is attended and assisted. If the nurse in attendance believes the problem may be portal vein thrombosis, and Watson provides a diagnosis of Marfan syndrome, the attending nurse can most certainly alert a physician to the suspected misdiagnosis.

Sure, that's how it will start. But after time goes by and many budget cuts later, the computer will be working alone. Okay, color me skeptic.

Everyday in every field of endevor humans do things that are spectacularly wrong. I think we should stop using them.

Indeed

^{_{{inspired by usbport}}}

I liked Jon Stewart's comment to Brian Christian while they discussed Watson and computer "evolution"...Stewart said something like..."so when a robot is walking down the street and trips and falls and looks up and says, 'I meant to do that.', then they will be advanced." I think that's about right.

Special purpose chips, computers etc. that are in "vital situations", will always be dependent on someone to look after them or at least check the results before a final decision on life and death is made by the human. Just like today you get a second opinion!

One of the things I find interesting about the latest "super computers", is how long they can stay running. A problem that always plagued us while building massive machines that required many thousands of parts, was the MTBF (Mean Time Between Failure). A big machine is no better than it's weakest component.