Emmanuel Maggiori - Get Tech: Understand Computers, the Internet and Cut Through the AI Hype. A crash course for non-techie entrepreneurs.

Get

Tech

Understand computers, the Internet and

cut through the AI hype

Emmanuel Maggiori

Contents

Introduction

I f you would you like to understand the world of computers, this book is written for you. Perhaps youre an entrepreneur, a business manager, a recruiter, or someone who likes to know how things work.

In your job, you may have to collaborate with programmers, AI specialists and other sorts of techies, but you may find it hard to communicate with them. They keep using cryptic words and getting caught up on problems you thought were easy to solve.

Or you may be an entrepreneur with a promising start-up idea, but youre having a hard time gauging how difficult it would be to build your product, or if it is buildable at all. If this is the case, youre not alone. Over the past few years, Ive been approached by countless entrepreneurs who had innovative ideas but were unsure about their technical feasibility.

For example, a few years ago, an entrepreneur had been concocting a business idea and wanted some advice on a couple of extra features he had in mind. One of them was a piece of software that would automatically analyze satellite images to predict how trees would grow in a neighborhood. It would then assess if a grown tree would block a nice view from a property and estimate the resulting drop in the property value. Thinking this was easy to do, he was considering adding it as a secondary feature in a credit risk-assessment app. Much to his chagrin, I had to tell this entrepreneur that making predictions from satellite images was a rather challenging task. This minor product feature hed thought up turned out to be pushing the boundaries of innovation and would require a lot of work.

This kind of confusion is to be expected, since no ones born knowing how computers work. And it is often counterintuitive: What is easy for us humans to do, like identifying a person in a picture, is not as easy for computers, and what is straightforward to do with computers, like multiplying large numbers, is not easy for us.

In fact, to work with computers effectively, techies train for years and end up developing a whole new language and a new way of approaching problems. This book will help you understand both.

The first part of this book is about computing. It will help you understand the basis of how computers process tasks and how these tasks need to be expressed for a computer to process them. You will get an idea of how software developers spend their time. You will also see that some problems are much harder to solve than they seem at first glance.

But we cannot speak about computing without discussing the Internet. Im sure youre a user of it yourself, possibly from several devices at once. And youve probably heard of IP addresses, port numbers, VPNs and encryption, but you may not know exactly what these terms mean.

If youre a tech entrepreneur, there is a high chance that whatever youre building it will have something to do with the Internet. But are you still wondering exactly what a front-end developer does or why you have to pay a fat bill to a cloud computing provider?

The second part of this book will help you understand how the Internet works and the challenges behind building Web applications. It will demystify terms that are commonly used but not always understood.

Finally, what about artificial intelligence? Everyone seems to have an opinion about it. Some people think the latest developments in AI are as revolutionary as the Internet was some time ago. You may have been told that your job is at risk due to AI. Youll soon be replaced by robots, they say. But, at the same time, youve been told for years that autonomous cars will soon be roaming the streets, yet you dont see them anywhere. How do you cut through the noise?

If youre an entrepreneur, youve probably been told you have to say youre doing AIits a magic word to gain attention from investors. I even knew a company that kept hiring AI specialists, even though their product didnt involve AI, just to claim that they were doing cutting-edge innovation. Its employees werent very happy about that.

The third part of this book will help you understand how AI really works. You might have heard buzzwords like deep learning or AlphaGo , and here youll see what they are really about. Ill also share with you the limitations of AI, which Ive seen firsthand by working in the field. If youre thinking of using AI in your business, this will help you understand its capabilities and why some things may not turn out as you expected.

I decided to write this book after speaking with many entrepreneurs who wanted to learn about computers but thought it was too hard. The thought of computer science textbooks and online coding classes was daunting enough that it wasnt even worth a try, they said. This book is neither of those things. It seeks to help you understand the world of computers yet fit into your busy life.

Now here we go. Enjoy the ride!

Part I

Computing

The Power and Limits of Computing

W ouldnt it be great if there was a toy model of a computer that could help you understand what any computer can do? It turns out that such a model exists and is really easy to grasp. This model is a great starting point for understanding the world of tech and the seemingly impossible complexity of computers today.

Alan Turing was a computer scientist, best known for designing the device used by the Allies in World War II to crack the German secret codes. But, a few years before that, in 1936, Turing presented a famous model of a problem-solving device. This model, developed years before modern computers appeared, was not necessarily meant to be constructed as is but to help the world understand the power of computers. It is now known as the Turing machine.

Turing presented compelling evidence that his machine, practicalities aside, was capable of solving any computable problem. Computable problems are those that can be solved by following a limited number of steps and thus are great candidates to be processed automatically by a machine. For example, the division of two numbers is computable. In fact, you probably learned in school (and, if you are anything like me, have since forgotten) a technique to divide numbers by pen and paper. This sort of recipe, which involves a series of systematic steps, is formally known as an algorithm . If you can write an algorithm for a problem, then you can solve it using Turings prototype.

This is a remarkable property because it sets the boundaries of computation: computers may come in all sorts of fancy packaging, but they cant be more capable than a Turing machine. Nearly a century later, no one has found a computable problem that cannot be solved by Turings prototype.

As you can imagine, modern computers are a bit different from Turing machines in their implementation, but they are fundamentally equivalent: today, an algorithm is coded in a programming language, but each of the elements in the program could be implemented in a Turing machine. Now lets see how a Turing machine works; this will help you understand what a computer can do, how it does it, and why a techie needs to approach a problem in a certain way to solve it with a computer.

Assume for now that youre given a very simple task, which goes as follows: You read inputs from a QWERTY keyboard. Every time the user presses the a key, a light automatically turns on and stays on until a different key is pressed. For example, if you type the word computational, the light bulb turns on (and off) twice, once for each a.

This machine would need to store information about which state it is in, in order to identify whether or not it is appropriate to turn on the light. You can see this kind of stateful behavior all around you on a daily basis. For example, a traffic light is at a certain state at any point in time: green, yellow or red. If a traffic light didnt know its current state, it wouldnt be able to decide which light to turn on. In your problem, the machine would need two states: one in which a has been pressed and the light should be on, and another state in which another key has been pressed and the light should be off.