Learning the terms used in construction is key to understanding how technology fits into the construction industry, and how that industry shapes the world.
- So let's start with some quick definitions. It really helps to use some common language, and to really understand what these terms mean because in all of my travels, I've, I continue to see people who say they understand terms and computation and really, really don't when I press them on the issue. I'll start with Cloud. Now (chuckles) I will admit, personally, I'm not a huge fan of the phrase Cloud, because we've really been in this yo-yo game in computation between on-site computing and off-site computing.
On-site mainframe and decentralized work stations. And originally, computers were very large mainframe and we had mainframe computers the size of this room that had a tenth of the computing capacity or a hundredth of the computing capacity of a phone. And then we moved to the personal computer, and we decentralized everything, and then we reconnected everything using networks. And so you've seen this continual ebb and flow of computing going from centralized to decentralized, but now, we're in this fascinating time in computational history, when we have both incredible power at the data center, incredible power at the workstation, and incredible connectivity between the two.
And that's allowing an array of options that have never been possible before, or really even dreamed of. So when we say Cloud, here's what we're really talking about. We're talking about scalable, self service, resource pooling, programmable, broad access. Now that's a bunch of nerdy phrases, so let me break it down for you in a little bit more basic terminology. It's computing that's available anywhere, anytime, on any device, and you can have as much as you want, whenever you want.
It's like the recipe for a great buffet, I mean, can you imagine a buffet of steak that was available anywhere, anytime and you can have as much as you want, whenever you want? That's what this is like, except it's for computing, storage, computation, processing. It's only made available because of all of these things converging simultaneously in the last few years. That's our definition of Cloud. Now there's three types of Cloud deployments that really matter, especially in construction.
There's private Cloud, which is when you have your own servers that are scalable enough to be considered Cloud. There's public Cloud, and that's really dominated the big, heavy there is Amazon with their Amazon web service offering, followed by Microsoft Azure, followed by Rackspace, followed by Google, followed, you can keep going down the list of Cloud service providers. And then they have a deployment that's really my favorite. Hybrid Cloud deployment. What hybrid Cloud deployment means is that you have local computing, Cloud-based resources at the job site.
And then you have server-based computing resources off-site, and the two synchronize. This is really useful because even today, with all of the great connectivity we have, we still have, continue to have connectivity problems out in the construction job site. Now this is a whole concept. Very, very important, and we still find professionals in the construction industry struggling and grappling with moving to a Cloud-based deployment, because they start with this fundamentally flawed assumption that computing that's here is safer than computing that's there.
Somehow, by locating the computing on-site, they feel like they're gaining additional control and security, but that's not the reality. The reality is that security-wise, if your computer is connected to the internet, or if it even has a USB port, you're just as vulnerable. So I need everybody to get over their hurts and hangups relating to local computing and recognize there's significant benefit to connecting a multi-hundred job site construction company, or even a four job site construction company together with Cloud-based computing.
We need to get over these flawed-based assumptions, because the reality is that there are security breaches every day at companies that don't use Cloud computing at all. You know, in fact, the only way to really secure your computer, by the way, rip the modem out, rip the wifi card out, rip the Bluetooth out, and rip all the USB ports out, and then only type things in and then read them off the screen. Now you know that's ridiculous. You could be a Luddite, and if you don't know what a Luddite is, go Google it.
It'll be an interesting read. You could be a Luddite and say, "You know what? "Forget this, "I'm going to use paper." I'm not really going to address that perspective (chuckles) in this video. But we, I have to address the fact that many people still have a hangup over migrating to Cloud-based computing. There's enormous benefits and enormous advantages. Let's move on to Big Data. Now I like Michael Stonebraker's definition from MIT of Big Data. Big volume, big velocity, big variety.
He really hit the nail on the head. Big Data is not the ERP system or accounting system in your company. Big Data is terabytes and terabytes of terabytes of information. Moving incredibly quickly from a bunch of disparate sources. I love the analogy of Niagara Falls. It's a bunch of water, and I mean a lot of water moving very, very quickly from a lot of different tributaries that combine into that river that falls over a cliff.
Except in a Big Data system in the Niagara Falls analogy, we could tell you where the water's coming from, where it's moving to, but most importantly, we can tell you what it all means. You see, Big Data gives meaning to seemingly meaningless piles of information. Now what is Big Data? I mean, you probably use Facebook or Twitter. That is big data. Terabytes per second of information flowing in. Video, photos, text, all unstructured, but they tell a story.
Did you know that we could actually determine the sentiment of the entire country by simply measuring the flow of data across Facebook and doing a sentiment analysis of the photos and text that flow across them? We can tell if America's having a good day or a bad day. Big Data is extremely powerful. It has huge implications in predicting material shortages, material pricing, labor supply, labor shortages, all of which can be predicted successfully with Big Data.
And that brings us to machine learning, 'cause we really have to have some fantastic computational capabilities to be able to use all of this data. Now machine learning is a form of AI. Artificial intelligence, whereby machines can be taught to learn. There are many cases when it's simply prohibited in both time and cost to program explicitly a machine to do all of the tasks that you need it to perform. And so we're looking at machine learning.
I really feel like it's good to go back to 1959. Arthur Lee Samuel, one of the granddaddies of computer science and really, one of the granddaddies of machine learning in specific AI. He wanted to teach a computer how to play checkers. Now I want you to think about checkers, and might not have played in a while. I suggest you go back. It's not as simple as it first seems because there's a lot of moves and countermoves that can be had. But what happens when you get to the end? You get kinged and you can move backwards, so there's really a complexity of moves.
He knew, using his computer in 1959, which, by the way, was a challenging computer to write code on, that it would take him prohibitively long time to program all the potential moves, so what did he do? He taught the machine to play itself. Over and over and over and over again. Thousands, and then tens of thousands, and then hundreds of thousands of times the machine played itself in checkers until it learned a reasonable probability matrix of how to win a game.
He had it play the world checkers champion and it won. This was a significant event. It was a significant event because for the first time, we were able to teach a machine how to actually teach itself. Now this is a form of specific AI. General artificial intelligence is a much more complex topic that we really won't go into, and it has not been solved. In fact, I would argue we're at least a decade away from general AI.
That's where a machine can really think, learn, and reason like a human being, or even better than a human being. We're not there. But on specific AI, we have numerous examples that have come into play in the recent past. And by recent, I mean the last 50 years. So there's some really interesting examples. One of the ones I'd like you to go to where you can see the Cloud and Big Data and machine learning converging is Google Images.
Now certainly Google Images, they're indexing all of the photos in the whole world that they can get access to. And they, the title in the file name, or the metadata in the file is not enough to tell you what's going on in that picture. So they created a machine learning algorithm that uses Big Data, this giant repository of photos, gathered with the Cloud, right? So this is truly a Cloud, Big Data, machine learning combined together into a simple text box. Go and type. Just type soccer player wearing a green shirt, carrying a yellow ball into Google Image search.
It will show you pictures of soccer players wearing a green shirt, carrying a yellow ball. How did it know that? 'Cause it used a type of machine learning called imaged recognition. And it used Big Data and the Cloud to be able to actually learn what objects look like on its own. We can't possibly program every permutation of a soccer ball, or of a shirt, or of a color. There's shades of color, there's types of soccer balls, there's types of shirts.
So go to google.com/images. You can also go to a really interesting website called captionbot.ai. This is an AI project for Microsoft that allows you to upload any photo you want. Now remember, please remember, only upload appropriate photos to the internet, because this become property of Microsoft because as you upload photos, it will write a caption for what's going on in that photo for you. It's a really great example, a really good teaser for you to go on and experiment with a definition of AI, machine learning, Big Data, and the Cloud.
It's an exciting field that has a ton of implications that we'll continue to explore for decades to come.
Follow James Benham—the CEO of JBKnowledge, Inc.—as he explains how construction science and computer science are merging into one joint field of study. James shares essential terms that you need to know to speak intelligently about topics like the cloud and machine learning. Plus, he dives into topics like the Internet of Things, the evolution of drones, and 3D printing. To wrap up the course, he covers IT budgets, staffing, and investing in research and development.
- Learning about the origins of construction technology
- Reviewing essential construction tech terms
- Understanding the Internet of Things
- Reviewing the evolution of drones
- Learning about the 3D printing process
- Investing in IT
- Investing in research and development