Are the World’s IT Infrastructures Today Even Too Complex for IBMWatson? It’s All About Generating The “Right
Posted Jan 08 2014 12:19pm
I have written several posts about the IBMWatson technology and sure when can crunch huge data bases of information and zero in on what you are pursuing as far as information that will help you make a better decision, what’s not to like and find value. Machine learning technologies have some limitations too according to this article. This is not IBM, but here’s what Google did to me, as it had not “learned enough” yet and their algorithms decided my name was not “machine compliant” and basically when reading all of the definitions of the word “duck” it has not run across it as a surname. This is why the process of “learning enough” is important.
Sure we have all read the news about Watson diagnosing and finding buried information relating to cancer but we come back around to the word “context” and I think that plays a role here too as we try and get the machines to think like a human as best we can and there still some short comings there obviously, work in progress, so too soon in some areas for high levels of confidence. The data may all be accurate in what it delivers, but is it in context of what you want for results?
Sure I wrote about Citibank going to use the Watson resources and saw it as a bit of a competitive threat for competitors and it still could be but after reading this article again we seem to come back to “context”. Scroll on down to the footer of this blog and watch the first video named “Context is Everything” and see what you think. Siefe does a good job in explaining how we get duped today with quantitated justifications…and it’s alive and well out there as we do put too much stock in numbers at times and they need to be accurate and work with the real world.
WellPoint health insurance is also out there as an IBM reseller and from reading what it stated here, it’s going to be a while before we can really tackle the “context” issue, and I write about that almost every day myself. I used to write software and thus know the data mechanics as I had to work and rework tons of queries myself to make sure on a simple SQL basis that they were delivering the right information and that it filtered though the software in screens to give the appropriate context to users. That really is a very big deal and takes time so you can read here how the doctors are working to get the same type of results here as well, except they have a lot more data and query functionalities to worry about for sure.
When IBMWatson first came out, I thought it would be great for Congress to use it to query and get information they need to create laws and I still think so rather than seeing a bunch of “carpetbagged” analytics and everyone fighting over who’s numbers are correct as they could all start with the same numbers this way. Had a conversation on the web with a banker last year and we both agreed that half of the analytics folks invest in will be a waste of money as the value won’t be there, yeah, rare me agreeing with a banker.
Here’s a couple posts where I talked about potential use in Congress with using speech recognition as even the lowest common denominator in who we elect with technology skills could participate and not get left out. If we had smarter folks in Congress we probably wouldn’t see the huge number of filibusters as that’s what happens when folks kind of run scared and don’t understand current day methodologies and technology I feel. The process still represents a good tool and 100% improvement over what we see today in DC.
You see there are no “algorithms fairies” out there and data has to work with the real world so in reading this article Watson stated the confidence level at less than 32%, which doesn’t mean it didn’t get the data, but what it brought out were treatments that probably didn’t have enough information and data to really make a confident recommendation, and stuff like that will happen, there’s no magic. I think HHS has found out the hard way this year when it comes to the reality of the complexities we have and you can’t just have any old figurehead at the top anymore and the same thing goes for the SEC and a few other agencies as they get stuck behind the 8 ball and then all the drama queen news begins.
We have a big awakening here and the process of trying to “dummy down” consumers as we have seen with some marketing will end up crashing as we are bit a smarter today and we need some common sense and some lines drawn here with when the virtual worlds and real worlds need to come together in a better fashion. My Algo Duping page has some great videos that explain a lot of this , so sit back for a few hours and take in what folks smarter than me have taken time to create. BD
“The short order computer code kitchen burned down several years ago and there was no fire sale”…Medical Quack…
IBM Chief Executive Virginia "Ginni" Rometty has told executives she hopes Watson will generate $10 billion in annual revenue within 10 years, according to an October 2013 conference-call transcript reviewed by The Wall Street Journal. She set that target after the executive in charge of Watson said its business plan would bring in $1 billion of revenue a year by 2018. That would make Watson the fastest IBM business unit to reach the $1 billion milestone.
But Watson had total revenue of less than $100 million as of late October, according to the transcript. One of its first big projects, with the University of Texas M.D. Anderson Cancer Center, was "in a ditch" in early 2013, said Manoj Saxena, the executive overseeing Watson.
IBM executives still believe Watson could become one of the biggest innovations in the company's 103-year history, alongside the mainframe and personal computer. In a sign of Watson's potential, IBM plans a major announcement about the business Thursday, said a person familiar with the matter.
Watson's key distinction from other analytical software is its ability to "learn." Feed it medical cases, and Watson will rank possible treatments by "confidence score." During training, doctors tell Watson when it makes a bad recommendation, and the supercomputer learns from its mistakes.
IBM is developing versions of Watson that can match cancer patients to clinical drug trials or recommend an investment strategy after reviewing a customer's portfolio.
Watson is having more trouble solving real-life problems than "Jeopardy" questions, according to a review of internal IBM documents and interviews with Watson's first customers.
At his office, he pulled out an iPad and showed a screen from Watson that listed three potential treatments. Watson was less than 32% confident that any of them were correct. "Just like cancer, it is much more complex than we thought," Dr. Kris said.