Share and discuss this blog

Thursday, April 28, 2016

Could IBM stop lying about Watson already? I guess not

IBM needs to stop lying. It is getting hard to take. Today alone I saw two outrageous lies about how Watson will save us all.

Here is the first article:

Its headline is:

Big Data: Will We Soon No Longer Need Data Scientists?

As you can guess, the answer is we won’t because Watson.

IBM, for example, believes that it can offer a solution to the skills shortage in big data by cutting out the data scientists entirely and replacing (or supplementing) them with its Watson natural language analytics platform.

I want to keep this simple, so I will say what I was doing today. I didn’t sleep well last night because of a phenomenon called alcohol rebound. I only had 2 drinks, but I had them 2 hours before bedtime  and this caused a rebound at 2 am which kept me up for hours. This has only started to happen to me in the last year or two, so I Googled “alcohol rebound in old people”  and found a long list of articles none of which were any help. I could ask my doctor but I am guessing he hasn’t memorized the literature and doesn’t know the data. But Watson can do it right? Watson wouldn't even understand my question much less my needs and it would not be able to extrapolate from data that might or might not be there. To put this another way, Google can’t answer most of the questions I pose to it and Watson is no better.  Natural language processing is not very good yet, no matter what all the “AI” deep learning people say. Intelligent people are always better to talk to than any AI system we currently envision.

These days we have large life insurance company as a client for one of our data analytics courses. So I imagined a  question they might ask Watson. “What is is the worst policy we could write?” They might ask that. Would Watson even know what “worst” meant in this context? Would it understand all the parameters relevant to determining an answer? I assume this company’s data scientists could answer this while it is safe to assume that Watson wouldn’t even know what the question meant. But this doesn’t stop them from advertising more nonsense about Watson. I had had enough.

And then I saw this:

The headline is

“C” is for cognitive learning
IBM and Sesame Street collaborate to create the next generation of tailored learning tools. This new technology venture combines Sesame Street’s expertise in education and storytelling with IBM Watson technologies.

And what piece of brilliance will Watson bring to education?Apparently they are just hopping on the personalized learning bandwagon, which means we will teach the stuff we are making you learn by tutoring you to get better test scores when we see what answers you got wrong. So, Watson will change learning, or maybe not so much. Watson will help kids who can’t read well by seeing what words they have trouble with and helping the kids practice. I have news for IBM. People can already do that. Good parents and teachers always do that. Is IBM’s view of education that all kids will have everything they do analyzed and then shoved at them again in another form because Watson is good at analyzing data? 

The problem in education is simple enough folks. It is boring. It is irrelevant to the interests and needs of most kids. They don’t need to learn classical Greek, or ancient history. They should be encouraged to learn what they want to learn. Could we do something radical and ask kids what they to learn how to do and then them help then learn to do it? We could, but then if we submitted the answer to Watson it wouldn’t understand a thing the kids responded. (What would it do with “I want to be a fireman?”)

(As an aside, people who read me regularly know that I am a terrible typist. Apple’s Pages does automatic spell correct and is very bad at doing that. But today it corrected my misspelling of Watson to Satan on two different occasions. Maybe Apple’s AI is smarter than Watson’s.

1 comment:

Janet Kolodner said...

Ha!! (to the last paragraph). Amen to the rest.