I had an old neighbor. She was real nice. Had a small dog. She Would walk the dog several times a day. Turns out she was a factory worker. High School degree. Lost her job. Out of work for over two years. Had to sell her house and move back home up north. In her 50's. She died a few months later. She was a nice person.
Without a skill which is marketable, this is a very tough economy. Going back to school is expensive, difficult, and may not get you a job. But what are the choices. The factory worker jobs are disappearing. What are these people supposed to do for a living.
A lot of news about raising the minimum wage for workers. Sounds good. But what if you don't have a job. Tons of competition for low wage jobs. And those jobs are not easy. Lots are customer facing. Physically demanding. Many are resorting to multiple streams of low paying jobs. With no insurance. Or upward mobility.
Technology is getting smarter, cheaper and more available. It's much easier to invest in technology than human capital. For obvious reasons. And that seems to be the trend. Displacing many workers. Banished out to pasture with no chance of return.
Why is technology disrupting the average worker?
Algorithms. Which are mathematical or statistical formulas. Which can learn, remember and get smarter over time. They are precise, fast and efficient. But most of all, they scale.
Algorithms can scan huge volumes of data, which would take a smart person or team of smart people decades to gather and research and perhaps form a conclusion. Deep learning machines can accomplish this, it's actually their main function.
Unsupervised learning machines, which are fed information, run through a neural network, which activate triggers, which activate more triggers downstream, are taught over time. They look at the 1000's of scenarios and are able to predict with fair amount of accuracy, future outcomes based on similar patterns. And they become resident experts in their narrow field of topic.
But getting them to learn across domains is the real challenge. However, they can apply some of their existing knowledge into learning other topics. So they don't have to start from the ground up.
I could envision a world where Unsupervised self learning machines run 24 hours per day. Connected to a central hub. And each of these independent neural networks would be linked to other neural networks, forming a grid of super computers across the globe. And these things would have vast knowledge on many subjects. Constantly learning and re-learning and fine tuning their Algorithms.
Eventually, more inputs would be added, based on visual images, sounds, sights, smells, voices, scenarios, events. Imagine a world with unlimited data storage, processing power and information. The machine would run continiously, processing data into information, perhaps generate probablistic outcome weighted predictions for humans to interpret and run scenarios, to see possible downstream effects. This would require some smart people to build the system, tune it as well as interpret the output.
These machines could potentially tackle some of man's biggest problems, like weather patterns, diseases, economics, who's going to with this weeks big game, population patterns and many complex subjects which baffle us today.
It still follows the classical computer model we learned about in basic computer class. Input device, Processor, output. Only difference, we are no longer dependent on punch cards, or tape drives, or floppy discs, or keyboards. Everything could go into the processor, stored, evaluated, crunched, churned into predictable repeatable patterns.
And that machine would surpass the smartest human rather quickly. Without all those emotions to distort, biases which skew results. But the politics of the computers owners are still up in the air. As this super machine's owners would have to have humanity's best interest. For if not, we are building a weapon, not an encyclopedia of everything.
So I was sad my neighbor lady passed away. She was very nice. And her dog liked our dogs.
Thanks for reading~!
This blog post is in no way an attempt to steal other people's work. It's basically an conglomeration of notes from research I did...
I signed up for the Hortonworks Certified Associate exam last Thursday. Figured if I sign up, I'd have to take the test. And if I tak...
Saw a post today on Twitter, " Microsoft releases CNTK, its open source deep learning toolkit, on GitHub " This is big news. Be...