- 7/9/2013 12:42:07 PM
I try hard not to be an ayesayer or a naysayer since that is such a complicated thing to take a position on...and it often doesn't matter, because things end up where they end up. When I said what I said, I did intend to point to the fact that we (collectively) can sometimes be a little too trusting of results that we can't comprehend, and too eager to implement some things on a large scale that can be undesirable (for commercial and other political reasons). If I have to state a position it would be that a good level of scientific skepticism is in order as we do big data experiments and not abdicate responsibility when we push the big Analyze-this button, especially if the follow-on action is also automated. Certainly, as with any other technical augmentation of human capabilities (e.g. arms), we get benefits as well as issues, neither of which can be attributed to the machines themselves.
- 7/9/2013 9:59:59 AM
Hi Balaji. So you write, "This is the big bad data that human beings can't handle, nor can inspect the truth of any patterns that the machines arrive at with their bottom-up analysis." That sounds a bit ominous to me, casting automated analytics in the shadows. Did you intend that? Are you a naysayer, in other words?
- by SethBreedlove, Data Doctor
- 7/8/2013 7:49:13 PM
Part of what makes us human is that we ask so many questions. I think Big-Data analytics will help us ask better questions and to come up with hypothesis that will more likely later become working theories or facts.
What I find really amazing about Target, is that they were also able to predict due dates. How Target Figured Out A Teen Girl Was Pregnant Before Her
I guess, watch what is in your mail. You may be able to learn family secrets.
- 7/8/2013 7:06:49 PM
I agree. We will probably have to depend on our own resources for poetry for a while at least. There are probably decisions of the poetry-kind though that we will expect our machines to make, and it may not always be clear that the machines are not quite ready for those kinds of decisions. The Wall Street models that were the superstition at the root of the collapse were a bit jarring. Models are great except when they miss the mark when they are aimed at the real world.
- 7/8/2013 6:36:24 PM
Beth, I see different kinds of automation based on different levels of outsourcing of decision-making. And different implications and trade-offs.
There are simple rules that can be programmed in or set by the user (e.g. I always book on a certain airline, unless there is a price differential of $X). Some of the rules you mention in the travel scenarios are of this kind.
Then there are decisions that we cede somewhat to the machines to let them decide based on patterns of behavior they detect (e.g. mail out baby care product catalogs to women who have an 80% probability of being pregnant) - things that are not life-threatening and that make it more efficient for the decision maker (e.g. a retailer). This is where the decision maker and other affected parties may have divergent interests.
Then we have the realm of things that are vague and fuzzy and voluminous. This is the big bad data that human beings can't handle, nor can inspect the truth of any patterns that the machines arrive at with their bottom-up analysis. Unless of course the machines share the "why" behind their decision-making process, and which we may inspect if the why's are not too voluminous for our capacity or patience.
- by PredictableChaos, Data Doctor
- 7/8/2013 4:33:54 PM
Machines are capable of enormous strides. Don't underestimate them.
However, I firmly believe there will always be a role for people. And it won't be just to program and maintain the machines.
- 7/8/2013 2:23:36 PM
If the results are optimized and the automation saves me time, money, and effort... well, why not? But I would prefer the ability to override a decision if necessary.
- by Noreen Seebacher, Blogger
- 7/8/2013 2:09:39 PM
Liberty Street Robot Supply & Repair — which bills itself as the Midwest's leading robot supply company — has announced a new line of decorative propaganda posters coming this summer.
The posters herald the coming robot revolution with taglines like, "Robots: A Good Idea!" and "A Robot Doesn't Need To Be Big. It Just Needs To Be Unstoppable." Designed by Oliver Uberti, this poster series showcases the bold graphic style that has become LSRS&R's signature.
Liberty Street Robot Supply & Repair is the storefront of 826michigan, a nonprofit serving 2,500 young people in Southeast Michigan with free creative writing and tutoring services. All proceeds from the Robot Revolution poster series will benefit 826michigan.